- Open Access
Systematic review of the evidence sources applied to cost-effectiveness analyses for older women with primary breast cancer
Cost Effectiveness and Resource Allocation volume 20, Article number: 9 (2022)
To appraise the sources of evidence and methods to estimate input parameter values in decision-analytic model-based cost-effectiveness analyses of treatments for primary breast cancer (PBC) in older patients (≥ 70 years old).
Two electronic databases (Ovid Medline, Ovid EMBASE) were searched (inception until 5 September-2021) to identify model-based full economic evaluations of treatments for older women with PBC as part of their base-case target population or age-subgroup analysis. Data sources and methods to estimate four types of input parameters including health-related quality of life (HRQoL); natural history; treatment effect; resource use were extracted and appraised. Quality assessment was completed by reference to the Consolidated Health Economic Evaluation Reporting Standards.
Seven model-based economic evaluations were included (older patients as part of their base-case (n = 3) or subgroup (n = 4) analysis). Data from younger patients (< 70 years) were used frequently to estimate input parameters. Different methods were adopted to adjust these estimates for an older population (HRQoL: disutility multipliers, additive utility decrements; Natural history: calibration of absolute values, one-way sensitivity analyses; Treatment effect: observational data analysis, age-specific behavioural parameters, plausible scenario analyses; Resource use: matched control observational data analysis, age-dependent follow-up costs).
Improving estimated input parameters for older PBC patients will improve estimates of cost-effectiveness, decision uncertainty, and the value of further research. The methods reported in this review can inform future cost-effectiveness analyses to overcome data challenges for this population. A better understanding of the value of treatments for these patients will improve population health outcomes, clinical decision-making, and resource allocation decisions.
Around half of the deaths from cancer occur in patients older than 70 years of age . Breast cancer is the most prevalent cancer for females, and older patients may have different treatment goals than a younger population . The increased likelihood of long-term comorbidities and frailty in this older population may preclude conventional treatment strategies (such as first-line surgery or adjuvant chemotherapy) due to the increased risk of treatment-related adverse events compared with younger people . As a consequence, clinicians and decision-makers may be uncertain about the most appropriate way to manage these older patients . Health economic evidence can inform treatment recommendations for breast cancer in older patients by comparing the incremental cost and health outcomes associated with different strategies available for this population . However, robust evidence for the relative cost-effectiveness of the various treatment strategies observed in routine practice for older patients with breast cancer, including non-surgical intervention, is currently sparse.
Decision-analytic models are essential to produce this cost-effectiveness evidence by synthesising all relevant evidence and extrapolating expected cost and health outcomes over a lifetime time horizon . As a minimum, health states such as 'disease-free', 'recurrence' (or 'progressed disease'), and 'dead' have been used previously to develop the structure of decision-analytic models for breast cancer . This structural characterisation of disease is unlikely to vary by the age of diagnosis. However, there are few sources of evidence derived from older patients to populate the input parameter values of these decision-analytic models. The majority of randomised controlled trials (RCTs) of treatments for breast cancer, for example, have either excluded older patients due to their higher risk of morbidity and mortality or have recruited relatively low numbers of older patients . Therefore, in the absence of data from older patients to populate key input parameter values, indirect evidence sourced from younger patients may be used instead to help estimate the cost-effectiveness of different treatment strategies for primary breast cancer in an older population.
Potential challenges may arise by using indirect evidence from younger patients if there are systematic differences with older patients in, for example, resource use, health-related quality of life (HRQoL), the natural history of the disease, or treatment benefits and harms. Older patients with breast cancer may have more interactions with the health care system and consume greater quantities of health care resources post-treatment than younger patients because of their relatively higher likelihood of comorbidity and frailty. Similarly, age-related comorbidities may result in older patients having relatively lower health-state utility values than younger patients . The natural history of the disease may vary between younger and older patients if prognostic factors (such as endocrine receptor positivity) differ across age groups . The magnitude and duration of benefit or direct harm from treatment (for example, the severity of adverse events after receiving chemotherapy) will likely depend on frailty experienced to a greater extent by older patients than younger patients .
In light of these potential differences between older and younger patients with primary breast cancer, if data from younger patients are used to populate input parameter values to estimate the expected cost and health outcomes of treatment strategies for older patients, analysts and decision-makers will need to appraise whether these sources of evidence are appropriate for the target population of the economic evaluation . Inappropriate input parameter values may result in inaccurate cost-effectiveness estimates, decision uncertainty, and the value of undertaking further research for older patients. Therefore, this study aimed to appraise the sources of evidence and methods to estimate input parameter values in decision-analytic model-based cost-effectiveness analysis of treatments for primary breast cancer in older patients (≥ 70 years old). The results from this study were then used to inform recommendations to improve the estimates of key input parameters in future cost-effectiveness analyses of treatments for older patients with primary breast cancer.
This study reports a systematic review of published economic evaluations of treatments (including surgery and any adjuvant or non-adjuvant treatments) for older females (≥ 70 years old) with early-stage primary breast cancer following the principles of the Preferred Reporting Items for Systematic Reviews and Meta-analyses Extension for Systematic Reviews (PRISMA) guidance . This review focused on the methods used by the included economic evaluations to estimate four types of input parameters: (i) health-related quality of life (HRQoL), (ii) the natural history of the disease, (iii) the magnitude of relative treatment effects, and (iv) resource use.
The criteria for inclusion and exclusion in the systematic review were based on the PICO framework , i.e., Population (older women aged 70 years or more with early-stage primary breast cancer), Intervention (any treatment, including surgery with or without adjuvant therapy), Comparator (any therapy), Outcome (incremental cost and health outcomes), and Study design (full economic evaluation) (Table 1). A full economic evaluation is defined as "the comparative analysis of alternative courses of action in terms of both their costs and consequences" , including cost-effectiveness analyses (CEA), cost-utility analyses (CUA) and cost–benefit analyses (CBA) that used a decision-analytic model. Conference abstracts and manuscripts were written in a non-English language were excluded (Additional file 1: Appendix 1).
Information sources and search strategy
Ovid EMBASE® (1974 to 2021 Week 35) and Ovid Medline® (1964 to September 2021) were searched electronically from inception until September 2021. The search strategy (Additional file 1: Appendix 2) comprised disease-specific terms for early-stage primary breast cancer and terms to identify published economic evaluations according to the filters reported by the Centre for Reviews and Dissemination .
Study selection and data collection
The titles and abstracts identified by the search strategy were screened independently for relevance against the inclusion criteria by two investigators (YW and LCC). The full texts of eligible studies were further retrieved and reviewed independently by two investigators (YW and LCC) to finalise study selection. At the full-text review stage, the age of the target population for the base-case analysis and, if relevant, for any age-specific subgroup analyses was identified within each economic evaluation to determine whether the study was designed for patients who were at least 70 years old. Discrepancies were resolved through consultation with a third reviewer (SG) to make a final decision.
Data extraction comprised two stages. In the first stage, the following data were extracted from each economic evaluation by one author (YW): (1) study design (country; target population; strategies compared), (2) study characteristics (evaluation method, i.e., CEA or CUA; type of decision-analytic model; time horizon; perspective; health outcome measure used, and costs included), (3) evidence sources that were used to estimate four types of input parameter values (HRQoL; the natural history of the disease; relative treatment effect; and resource use/cost), (4) methods of analysis (whether deterministic/probabilistic sensitivity analyses or value of information (VOI) analyses were reported), and (5) estimated results (base-case and sensitivity analyses, VOI, and key drivers of relative cost-effectiveness through sensitivity analysis). In the second stage of data extraction, the characteristics of the estimation sample (sample size and mean age) were extracted from the original sources of evidence used by the included economic evaluations to estimate their input parameter values.
The completeness of reporting in each economic evaluation was assessed by 17 items in the Consolidated Health Economic Evaluation Reporting Standards (CHEERS) statement . Full adherence to any item was noted as 'Yes', partially adherence was indicated as 'Partial', and non-adherence as 'No'. Two researchers (YW and LCC) independently appraised each identified economic evaluations' quality. Any discrepancies were discussed with a third reviewer (SG) to make a final decision. Quality assessment was summarised visually and reported by a narrative synthesis.
The extracted data from each economic evaluation were first reported in a table and summarised by a narrative synthesis. This summary described the sample of included economic evaluations according to the type of decision-analytic model used, the proportion of studies that had a target population of patients at least 70 years old in either the base-case or subgroup analysis, the treatment strategies compared, and the main results of each economic evaluation. For each economic evaluation, the sources of evidence used to estimate four types of input parameter were then appraised to determine whether they were obtained from an estimation sample that corresponded with the age of the target population (i.e., ≥ 70 years old). For the remainder of this study, (1) 'HRQoL' refers to the health state utility values, (2) the 'natural history of disease' refers to the probability of health events in the absence of a treatment effect, (3) the 'relative treatment effect' refers to the magnitude of difference between two treatments, and (4) 'resource use' refers to the direct health care resources consumed by patients. In the cases where evidence for input parameter values was based on an estimation sample of patients aged less than 70 years old, the methods of each economic evaluation were then appraised to determine whether any adjustment or calibration was performed to make these estimated values more appropriate for an older population.
The PRISMA diagram (Fig. 1) illustrates the identification, screening and inclusion of studies. The electronic database searches identified 3544 studies, and 67 were read in full. The final sample comprised seven decision-analytic model-based economic evaluations of treatments for primary breast cancer in patients aged 70 years or more [18,19,20,21,22,23,24] (Fig. 1).
Characteristics of included studies
All of the included economic evaluations reported both CEA and CUA. The decision-analytic models used by the identified economic evaluations included a cohort Markov model (n = 3) [20, 21, 24] and a patient-level simulation (n = 2) [22, 23]. Two studies did not report the type of decision-analytic model [18, 19]. All the seven economic evaluations used at least three health states within the structure of their decision-analytic model (disease-free; progressed disease; and dead). Different clinical outcomes were used between the economic evaluations to define the health state for progressed disease, including recurrence, local relapse, or metastasis. The structure of the decision-analytic model in four studies [21,22,23,24] also included an additional health state for treatment side effects (Table 2, Full data extraction in Additional file 1: Appendix 3).
Three economic evaluations (43%) had a base-case target population that focused exclusively on older patients aged ≥ 70 years [20, 22, 23]. The two studies by Ward et al. [22, 23] had a base-case target population of patients aged 70 years or older with estrogen-positive invasive breast cancer. Sen et al.  reported results for 70 years, 75-years, and 85-years old with early-stage breast cancer. Four economic evaluations (57%) reported cost-effectiveness estimates for older patients as part of subgroup analysis by age [18, 19, 21, 24]. The two studies by Naeim et al. reported results for subgroups of patients aged 75-years and 85-years old who had early-stage node-positive  and node-negative  breast cancer. Desch et al.  reported results for a subgroup of patients aged 60-years to 80-years old with a diagnosis of primary breast cancer, and Skedgel et al.  reported results for subgroups of patients aged 70 years and ≥ 80-years old.
Three studies [20, 21, 24] compared surgery alone with either adjuvant chemotherapy alone , radiotherapy , or chemotherapy ± trastuzumab . The results from these three studies indicated that surgery alone was more cost-effective than surgery plus adjuvant treatments for the older population [20, 21, 24] (Table 2). Two studies compared surgery plus adjuvant chemotherapy with adjuvant chemotherapy ± endocrine therapy [18, 19]. Two studies compared surgery plus adjuvant radiotherapy with adjuvant endocrine therapy and their combination [22, 23]. Of these four studies, which compared different adjuvant strategies, the estimated results suggested that less adjuvant treatment, or less harmful adjuvant treatment (i.e., less intensive radiotherapy or less toxic chemotherapy), was more cost-effective for older patients with breast cancer [18, 19, 22, 23]. No published economic evaluation compared surgery with non-surgical treatment as the initial strategy to manage older patients with primary breast cancer (Table 2). In addition, no identified economic evaluation reported a value of information (VOI) analysis to investigate the need for further research to reduce uncertainty in the estimates of relative cost-effectiveness  (Table 2).
Table 3 reports the quality assessment of the seven economic evaluations according to the CHEERS criteria (Table 3). Thirteen domains of the CHEERS criteria (76%) were reported clearly by the included studies. However, in general, the economic evaluations whose base-case target population comprised older patients exclusively reported the sources of evidence to estimate input parameters more clearly than the economic evaluations that reported results for older patients as part of a subgroup analysis (Table 3). Six economic evaluations partially reported their analytical methods and study parameters which justifies the critical appraisal of these values for the remainder of this review.
Analysis of evidence sources for input parameters
The economic evaluations' sources of evidence and methods to estimate four types of input parameters (HRQoL, natural history, treatment effect, and resource use) are reported below. (Details of Input Parameters in Additional file 1: Appendix 4).
Health-related quality of life
All seven economic evaluations reported expected health outcomes as quality-adjusted life years (QALYs) [18,19,20,21,22,23,24]. The EQ-5D instrument was used to estimate HRQoL in four studies [20,21,22,23]. Three studies estimated HRQoL values by expert elicitation [18, 19, 24]. Across the included economic evaluations, four approaches were taken to make the HRQoL values a function of the target population's age: (1) HRQoL, which was independent of age; (2) partial age-dependent HRQoL; (3) age-dependent HRQoL with a disutility multiplier; and (4) age-dependent HRQoL with an additive utility decrement (Table 4).
The two studies by Naiem [18, 19] used HRQoL values fixed across age subgroups and independent of the target population's age. Patients were assumed to have lower utility if they received hormone therapy (HRQoL = 0.99), or chemotherapy with minor toxicity (HRQoL = 0.90) or major toxicity (HRQoL = 0.8). Similarly, Desch  also assumed that patients had the same utility values after experiencing minor and major side-effects from chemotherapy. This approach may overestimate the expected QALYs accrued by older patients if the loss of HRQoL due to treatment-related adverse events is greater than for younger patients.
Skedgel et al.  estimated HRQoL values, which were partially dependent on the age of the target population. The utility values for patients who were 'disease free' were calculated using EQ-5D data from the Medical Expenditure Panel Survey (MEPS) between 2000 and 2002  (n = 38,678 adults). This approach enabled the authors to account for the negative association between age and HRQoL in the general population. However, the HRQoL values for subsequent health states (e.g., recurrence, second recurrence) and adverse events (e.g., nausea) appeared to be fixed and independent of age.
Sen et al.  used a disutility multiplier to estimate HRQoL values, which depended on the age of the target population. The MEPS (1998–99) was also used by Sen et al. to estimate age-dependent EQ-5D values for patients after successful treatment to preserve the negative association between age and HRQoL in the general population. Utility values for subsequent health states (e.g., local recurrence) were estimated from a published standard gamble study with 97 patients . The authors then adjusted these utility values using a disutility multiplier based on the mean age-dependent EQ-5D values from the MEPS. This approach ensured that, on average, the HRQoL values accrued by patients who experienced these subsequent health states reflected the observed decline of HRQoL over their lifetimes. For example, the estimated HRQoL value for local recurrence was, therefore, lower for older patients than for younger patients.
Ward et al. [22, 23] used an additive utility decrement to estimate HRQoL values, which depended on the patient's age. A representative cross-sectional survey of the US population (n = 4,000) estimated a baseline EQ-5D value for 70 year old females between 2005 and 2006 . The majority of subsequent health states had a corresponding disutility which was subtracted from this baseline EQ-5D value as an additive decrement (i.e., baseline HRQoL – disutility = new HRQoL). Similar to Sen et al., this approach enabled the authors to estimate HRQoL values for patients who entered subsequent health states, which accounted for the lower HRQoL experienced by older patients, on average, compared with younger patients.
Natural history of the disease
The included economic evaluations used four methods to estimate input parameters that reflected the natural history of breast cancer: (1) data were used from younger patients without adjustment; (2) data were used from older patients without adjustment; (3) plausible values were assumed and varied in a sensitivity analysis, and (4) data were used from younger patients and calibrated for an older population (Table 5).
The two economic evaluations by Naiem et al. [18, 19] estimated the 10-year breast cancer-specific mortality for patients aged 75-years and 85-years old from studies where the estimation sample was younger (e.g., between 50 and 55% of the sample was below 55-years old). This approach may have underestimated the probability of death in the target population if the 10-year breast cancer-specific mortality is higher for older patients than younger patients. By contrast, Sen et al.  estimated the probability that patients experience health states (e.g., recurrence and metastasis) from published data of a trial that had recruited a sample of older females with breast cancer. It is likely that the probabilities derived from these trial data were more representative of an older population, given that the target population of Sen et al.'s economic evaluation and the estimation sample of the trial had similar patient characteristics.
Skedgel et al. argued that the prior probability of recurrence was unknown for their target population. The authors assumed a range of plausible values for the probability of recurrence across different ages to handle this. One advantage of this approach was that the impact of varying the probability of recurrence on cost-effectiveness estimates could be explored in a sensitivity analysis. Ward et al. [22, 23] estimated transition probabilities using data from a published trial that had whose sample was younger than 70 years old [23, 29,30,31,32,33,34,35]. To make these data more representative of an older population, the authors used calibration methods by applying a 'reduction factor' to the annual event rate in both arms of the trial. This approach reduced the absolute risk of events and made the input parameter values more appropriate for an older population.
Magnitude of treatment effects
The economic evaluations used four methods to incorporate age-specific heterogeneity in the relative treatment effects. These methods include (1) direct estimation of age-specific treatment effects from RCT or meta-analysis data; (2) scenario analyses of plausible age-specific treatment effects in the absence of data; (3) the use of observational patient-level data to estimate age-specific treatment effects; and (4) the incorporation of age-specific behavioural parameters to modify the treatment effect.
Skedgel et al.  assumed that the relative treatment effect for adjuvant chemotherapy was a function of the patient's age. The authors estimated hazard ratios for 'premenopausal' (40 and 50 years) and 'postmenopausal' (60 and 70 years) patients using RCT data [36, 37]. These data indicated, for example, that adjuvant chemotherapy was less effective at reducing recurrence for older patients (HR: 0.672) than younger patients (HR: 0.563). The authors then assumed that the relative treatment effect for adjuvant trastuzumab was the same for both older and younger patients. Similarly, Desch et al.  assumed that the annual relative reduction in recurrence for patients aged 60 to 69-years old was 20%, compared with 30% for younger patients, according to data from a meta-analysis of RCTs [38, 39].
Naiem et al. [18, 19] first estimated the relative treatment effect of adjuvant therapies (odds-reduction of 10-year mortality) from a meta-analysis of RCTs for patients aged 45-years and 65-years old [40,41,42,43]. In the absence of evidence for the relative treatment effect in 75-years and 85-years old patients, the authors assumed three possible values (low, medium, and high) of treatment effects. In the 'high' scenario, the magnitude of the treatment effect was assumed to be equivalent to that for a 65-year old patient. The authors then estimated how reducing this treatment effect in older patients may impact cost-effectiveness estimates by using the 'medium' and 'low' scenario analyses. The details to calculate the medium values (extrapolated the trend of less benefit with increasing age) and the low values (minimal benefit) were not described explicitly.
Sen et al.  incorporated age-specific heterogeneity in the relative treatment effect by performing a patient-level analysis of data from the observational Surveillance, Epidemiology, and End Results (SEER) Program . The authors estimated the 5-year and 10-year overall survival from radiotherapy compared with surgery. The estimated treatment effects were stratified by three age groups (70–74; 75–79; and 80–89 years old).
Ward et al. [22, 23] incorporated a behavioural parameter to reflect evidence that adherence to endocrine therapy may reduce in older patients. Data from a registry study of patients at least 65-years old estimated that compliance with endocrine therapy was 61% at 5-years. In the economic evaluation, this reduction of adherence had a subsequent impact on the relative effectiveness of endocrine therapy. By including this behavioural parameter, the authors were able to model potential changes in the relative effectiveness of treatment as patients became older.
Resources and cost
The included economic evaluations used two methods to estimate input parameters for resource use: (1) estimated input parameters were independent of age, and (2) estimated input parameters were dependent on age (Table 6).
Five economic evaluations assumed that estimates of resource use were independent of each patient's age. Naeim and Keeler [18, 19] estimated the resource use for managing side effects of adjuvant chemotherapy (10% of patients needed treatment to manage low white cell counts, and 3% of patients required hospitalisation for neutropenic fever) based on data from an RCT that had a sample of younger patients (81% of the sample was ≤ 49-years old) . However, this approach may have underestimated the resources required if hospitalisation rates or treatment for low white cell counts are higher in an older population . Skedgel et al.  extracted the local and distant recurrence costs from published costing study . These cost estimates were fixed for all age subgroups. The mean age of the sample in the published costing study was not reported, so it was not clear whether these data were applicable for a population of 70 year-old patients with primary breast cancer. Ward [22, 23] estimated direct and indirect costs using a hospital database and clinical guidelines. However, the authors did not report how the estimated cost of the metastatic disease ($23,460) was calculated.
Two economic evaluations estimated age-specific input parameter values for resource use [20, 24]. Desch  extracted cost data from the previously published economic evaluations [48, 49] and assumed that total costs of breast cancer treatment decreased as patients got older. This assumption was based on the reduction in follow-up costs, fewer late recurrences, and increased mortality from other causes over time. Sen estimated age-specific (70–74, 75–79, and 80–94 years old at diagnosis) cancer-related costs by conducting a matched cohort study from the SEER-Medicare database. Cancer patients were matched with non-cancer patients based on age, race, comorbidity, region, and year of diagnosis. All costs (inpatient, outpatient, physician, home health, hospice, and Durable Medical Equipment claims) for cancer and non-cancer patients were estimated over the 2-months before and up to 12-months after, date of diagnosis, and then stratified by type of initial treatment received. The cancer-related costs were the difference between the total cost accrued by cancer patients and their matched control .
Decision-analytic model-based economic evaluations will be essential to help inform the growing interest from decision-makers and clinicians about how best to treat older patients diagnosed with primary breast cancer. However, this review found just seven economic evaluations of treatments for this older population, and all studies compared adjuvant strategies only [18,19,20,21,22,23,24]. The authors of these economic evaluations used different methods to estimate input parameters values for HRQoL, the natural history of breast cancer, relative treatment effects, and resource use by using data from both older and younger patient populations. Therefore, a gap exists between the economic evidence required by decision-makers and the economic evidence available currently for managing primary breast cancer in older patients. To help close this evidence gap, the different methods to estimate age-specific input parameters reported in this review can inform the design of future model-based economic evaluations and strategies to overcome the relative scarcity of data from older patients.
A key distinction between the identified economic evaluations was whether they reported cost-effectiveness evidence for older patients as the base-case analysis or as part of subgroup analysis. For example, over half (57%) of economic evaluations in this review reported a subgroup analysis for patients older than 70 years old. Subgroup analyses in economic evaluations are a valuable method to investigate heterogeneity in cost-effectiveness by identifiable patient characteristics . However, it is vital to ensure that input parameter values are appropriate for each subgroup under investigation. If future economic evaluations report age-specific subgroup estimates for older patients with primary breast cancer, decision-makers and analysts should appraise whether the input parameter values are expected to vary across age groups or whether they are independent of age. This will improve the face validity of the model-based analysis and the external validity of the subgroup estimates of cost-effectiveness.
The National Institute for Health and Care Excellence (NICE) Decision Support Unit and the International Society for Pharmaceutical Outcomes Research advise that HRQoL should be estimated using evidence from a population that is similar (e.g., age, sex, and disease severity) to the modelled population [51, 52]. Age is a key determinant of HRQoL because older patients may have lower values than younger patients due to comorbidities and frailty . In this review, the methods used by Ward et al. [22, 23] and Sen et al.  to estimate HRQoL (i.e. disutility multipliers or additive utility decrements informed by baseline values from representative surveys of the general population) are helpful techniques for future economic evaluations to incorporate age-specific input parameter values when only data from younger patients are available. Peasgood et al.  report a systematic review and meta-regression of health state utility values for breast cancer. Many economic evaluations have previously obtained relevant input parameter values from this study. This meta-regression could be developed further by investigating whether including the mean age of the patients in each study affects the estimated relationship between health state utility and the other variables. Similarly, some studies have used mapping methods to estimate the statistical association between breast cancer-specific patient-reported outcomes (e.g., the European Organization for Research and Treatment of Cancer Quality of Life Questionnaire, EORTC QLQ-C30 ) and generic instruments used to estimate health state utility values (e.g., EQ-5D) [56,57,58]. Future mapping studies could build on these earlier studies by including an age variable to account for the impact of older age on HRQoL.
Depending on the decision problem, older patients' natural history of breast cancer may be available to estimate transition probabilities between health states. However, most economic evaluations in this review used evidence from a younger population to estimate these natural history parameters. The calibration method by Ward et al. [22, 23], which adjusted estimates from younger patients so that they were appropriate for an older population, is a helpful technique for future economic evaluations when data from patients over 70 years old are not available. Alternatively, future economic evaluations could use formal expert elicitation methods  to estimate these natural history parameters with sensitivity analyses around plausible values. There has also been an increase in the availability of linked primary care, secondary care, mortality, and cancer register data sources . These linked data could also be a valuable source of evidence to estimate the natural history of breast cancer for older patients in routine practice, ensuring that any selection bias and confounding are accounted for.
The magnitude of the estimated relative treatment effects and toxicity for patients with breast cancer can vary by age and type of treatment . For example, chemotherapy and radiotherapy have shown a limited survival benefit for older patients with primary breast cancer than their younger counterparts . Similarly, the effectiveness of endocrine and biological therapy is highly associated with the level and sensitivity of hormone receptors and HER-2 receptors. Older patients have a higher level of ER and PR receptors, whereas a lower level of HER-2 receptors [62, 63]. For future economic evaluations, the target population should be defined clearly in terms of whether a patients' age interacts with the biological mechanisms of disease and, as a consequence, whether the estimated treatment effects are appropriate for that population. The choice of comparator strategies should also be limited to relevant ones for the target population in routine practice. For example, non-surgical strategies may be the most appropriate comparators for patients who are ineligible for surgery due to frailty.
Older patients consume more health care resources than younger patients [63, 64] and costing studies from the US and UK [65, 66] indicated that the main cost drivers for cancer treatment in older populations were from treating side effects and related health care (for example, care and management of chemotherapy-induced neutropenia, radiotherapy-induced skin/gastrointestinal reaction, and trastuzumab induced cardiotoxicity). It is essential for future economic evaluations of treatments for primary breast cancer to report the evidence sources for resource use transparently to help decision-makers appraise whether these data can generalise to an older population. Sen  undertook a matched cohort study to estimate the incremental cost of managing older patients with primary breast cancer. Future research could use a matched cohort design to estimate valuable resource use data for older patients with primary breast cancer using large national observational datasets which link secondary care resource use with cancer diagnosis data. These patient-level data could then provide a better characterisation of how parameter uncertainty in estimates of resource use is distributed.
One limitation of this review was that the search strategy only identified published economic evaluations from peer-reviewed academic journals and may have missed some economic evaluations in the grey literature from government or private organisations. However, the sample of included studies successfully identified a broad range of different methods used to estimate input parameter values for an older population. A second potential limitation was that this systematic review focused only on four specific input parameter types. Therefore, valuable methods to estimate other input parameter types may have been omitted. However, the focus on input parameters for HRQoL, the natural history of the disease, treatment effects, and resource use was sufficient to characterise the majority of essential input parameters for any model-based cost-effectiveness analysis.
Future research could begin to estimate the cost-effectiveness of the different strategies along the full pathway of care observed in routine practice to manage older patients with breast cancer. This economic evidence would be valuable to inform how best to treat these patients by simultaneously considering health outcomes and costs to the healthcare system. Future research could also undertake a value of information analysis, based on the probabilistic outputs from these model-based analyses, to establish whether subsequent primary research in older patients would be worthwhile to reduce uncertainty in estimates of cost-effectiveness. Finally, future research could appraise the sources of evidence and methods to estimate input parameter values within economic evaluations of treatments for older patients diagnosed with other types of primary cancer.
The number of patients older than 70 years of age diagnosed with primary breast cancer is increasing. Health economic evidence will be essential to inform how best to manage these patients. This systematic review found only seven CEAs for this older population, indicating that further economic evidence will be valuable to meet the needs of decision-makers and service commissioners in the future. The methods to estimate input parameters described in this systematic review can help analysts overcome common data challenges to improve the accuracy of expected cost and health outcome estimates. Well-designed observational studies using national register data and formal expert elicitation exercises also present a considerable opportunity to improve the quality of input parameters estimates for this older patient population. A greater emphasis on understanding the cost-effectiveness of care for older patients with primary breast cancer will simultaneously improve population health outcomes, clinical decision-making for these patients, and the allocation of limited resources for health care.
Availability of data and materials
Supplement information is included in the appendices.
University of Oxfrod GCDL. Our World in Data. 2021 [cited 2021 06 Sep]; https://ourworldindata.org.
Tang SW, et al. Is surgery always indicated in older women with breast cancer? Minerva Chir. 2010;65(5):555–68.
Given B, Given CW. Older adults and cancer treatment. Cancer. 2008;113(12 Suppl):3505–11.
Extermann M, et al. Priorities for the global advancement of care for older adults with cancer: an update of the International Society of Geriatric Oncology Priorities Initiative. Lancet Oncol. 2021;22(1):e29–36.
Drummond MF, et al. Methods for the economic evaluation of health care programmes. Oxford University Press; 2015.
Brennan A, Akehurst R. Modelling in health economic evaluation. What is its place? What is its value? Pharmacoeconomics. 2000;17(5):445–59.
Annemans L. Methodological issues in evaluating cost effectiveness of adjuvant aromatase inhibitors in early breast cancer: a need for improved modelling to aid decision making. Pharmacoeconomics. 2008;26(5):409–23.
Parks RM, Holmes HM, Cheung KL. Current challenges faced by cancer clinical trials in addressing the problem of under-representation of older adults: a narrative review. Oncol Ther. 2021;9(1):55–67.
Ara R, Brazier JE. Populating an economic model with health state utility values: moving toward better practice. Value Health. 2010;13(5):509–18.
Parks RM, Green AR, Cheung K-L. The impact of tumour biology on the management of primary breast cancer in older women—based on a research programme in Nottingham. Ann Breast Surg. 2021;5:5.
Kojima G, Liljas AEM, Iliffe S. Frailty syndrome: implications and challenges for health care policy. Risk Manag Healthc Policy. 2019;12:23–30.
Kaltenthaler E, Tappenden P, Paisley S. Reviewing the evidence to inform the population of cost-effectiveness models within health technology assessments. Value Health. 2013;16(5):830–6.
Page MJ, et al. The PRISMA 2020 statement: an updated guideline for reporting systematic reviews. BMJ. 2021;372:n71.
Methley AM, et al. PICO, PICOS and SPIDER: a comparison study of specificity and sensitivity in three search tools for qualitative systematic reviews. BMC Health Serv Res. 2014;14:579.
Rudmik L, Drummond M. Health economic evaluation: important principles and methodology. Laryngoscope. 2013;123(6):1341–7.
Dissemination, C.f.R.a. Systematic reviews: CRD’s guidance for undertaking reviews in health care. 1st ed. Centre for Reviews and Dissemination: York; 2009.
Husereau D, et al. Consolidated Health Economic Evaluation Reporting Standards (CHEERS)–explanation and elaboration: a report of the ISPOR Health Economic Evaluation Publication Guidelines Good Reporting Practices Task Force. Value Health. 2013;16(2):231–50.
Naeim A, Keeler EB. Is adjuvant therapy for older patients with node (+) early breast cancer cost-effective? Breast Cancer Res Treat. 2005;94(2):95–103.
Naeim A, Keeler EB. Is adjuvant therapy for older patients with node (-) early breast cancer cost-effective? Crit Rev Oncol Hematol. 2005;53(1):81–9.
Sen S, et al. Examining the cost-effectiveness of radiation therapy among older women with favorable-risk breast cancer. J Natl Cancer Inst. 2014;106(3):dju008.
Skedgel C, Rayson D, Younis T. Is adjuvant trastuzumab a cost-effective therapy for HER-2/neu-positive T1bN0 breast cancer? Ann Oncol. 2013;24(7):1834–40.
Ward MC, et al. Cost-effectiveness analysis of endocrine therapy alone versus partial-breast irradiation alone versus combined treatment for low-risk hormone-positive early-stage breast cancer in women aged 70 years or older. Breast Cancer Res Treat. 2020;182(2):355–65.
Ward MC, et al. Radiation therapy without hormone therapy for women age 70 or above with low-risk early breast cancer: a microsimulation. Int J Radiat Oncol Biol Phys. 2019;105(2):296–306.
Desch CE, et al. Should the elderly receive chemotherapy for node-negative breast cancer? A cost-effectiveness analysis examining total and active life-expectancy outcomes. J Clin Oncol. 1993;11(4):777–82.
Fenwick E, et al. Value of information analysis for research decisions-an introduction: report 1 of the ISPOR value of information analysis emerging good practices task force. Value Health. 2020;23(2):139–50.
Sullivan PW, Ghushchyan V. Preference-Based EQ-5D index scores for chronic conditions in the United States. Med Decis Making. 2006;26(4):410–20.
Hayman JA, et al. Patient preferences concerning the trade-off between the risks and benefits of routine radiation therapy after conservative surgery for early-stage breast cancer. J Clin Oncol. 1997;15(3):1252–60.
Fryback DG, et al. US norms for six generic health-related quality-of-life indexes from the National Health Measurement study. Med Care. 2007;45(12):1162–70.
Hughes KS, et al. Lumpectomy plus tamoxifen with or without irradiation in women age 70 years or older with early breast cancer: long-term follow-up of CALGB 9343. J Clin Oncol. 2013;31(19):2382.
Kunkler IH, et al. Breast-conserving surgery with or without irradiation in women aged 65 years or older with early breast cancer (PRIME II): a randomised controlled trial. Lancet Oncol. 2015;16(3):266–73.
Potter R, et al. Lumpectomy plus tamoxifen or anastrozole with or without whole breast irradiation in women with favorable early breast cancer. Int J Radiat Oncol Biol Phys. 2007;68(2):334–40.
Blamey RW, et al. Radiotherapy or tamoxifen after conserving surgery for breast cancers of excellent prognosis: British Association of Surgical Oncology (BASO) II trial. Eur J Cancer. 2013;49(10):2294–302.
Fisher B, et al. Tamoxifen, radiation therapy, or both for prevention of ipsilateral breast tumor recurrence after lumpectomy in women with invasive breast cancers of one centimeter or less. J Clin Oncol. 2002;20(20):4141–9.
Early Breast Cancer Trialists’ Collaborative Group. Aromatase inhibitors versus tamoxifen in early breast cancer: patient-level meta-analysis of the randomised trials. Lancet. 2015;386(10001):1341–52.
Early Breast Cancer Trialists’ Collaborative Group. Relevance of breast cancer hormone receptors and other factors to the efficacy of adjuvant tamoxifen: patient-level meta-analysis of randomised trials. Lancet. 2011;378(9793):771–84.
Smith I, et al. 2-year follow-up of trastuzumab after adjuvant chemotherapy in HER2-positive breast cancer: a randomised controlled trial. Lancet. 2007;369(9555):29–36.
Perez EA, et al. Four-year follow-up of trastuzumab plus adjuvant chemotherapy for operable human epidermal growth factor receptor 2–positive breast cancer: joint analysis of data from NCCTG N9831 and NSABP B-31. J Clin Oncol. 2011;29(25):3366.
Fisher B, et al. A randomized clinical trial evaluating sequential methotrexate and fluorouracil in the treatment of patients with node-negative breast cancer who have estrogen-receptor-negative tumors. N Engl J Med. 1989;320(8):473–8.
Rosen PR, et al. A long-term follow-up study of survival in stage I (T1N0M0) and stage II (T1N1M0) breast carcinoma. J Clin Oncol. 1989;7(3):355–66.
Peto R. Effects of adjuvant tamoxifen and of cytotoxic therapy on mortality in early breast cancer. An overview of 61 randomised trials among 28,896 women. Harmone Res Pediatr. 1989;32(Suppl1):165–165.
Early Breast Cancer Trialists’ Collaborative Group. Systemic treatment of early breast cancer by hormonal, cytotoxic, or immune therapy: 133 randomised trials involving 31 000 recurrences and 24 000 deaths among 75 000 women. Lancet. 1992;339(8784):1–15.
Early Breast Cancer Trialists’ Collaborative Group. Tamoxifen for early breast cancer: an overview of the randomised trials. Lancet. 1998;351(9114):1451–67.
Early Breast Cancer Trialists’ Collaborative Group. Polychemotherapy for early breast cancer: an overview of the randomised trials. Lancet. 1998;352(9132):930–42.
Warren JL, Klabunde CN, Schrag D, Bach PB, Riley GF. Overview of the SEER-Medicare data: content, research applications, and generalizability to the United States elderly population. Med care. 2002;40(Supplement):IV3–IV18. https://doi.org/10.1097/00005650-200208001-00002.
Fisher B, et al. Two months of doxorubicin-cyclophosphamide with and without interval reinduction therapy compared with 6 months of cyclophosphamide, methotrexate, and fluorouracil in positive-node breast cancer patients with tamoxifen-nonresponsive tumors: results from the National Surgical Adjuvant Breast and Bowel Project B-15. J Clin Oncol. 1990;8(9):1483–96.
Lyman GH, et al. Risk models for predicting chemotherapy-induced neutropenia. Oncologist. 2005;10(6):427–37.
Will BP, Berthelot J-M, Le Petit C, Tomiak EM, Verma S, Evans WK. Estimates of the lifetime costs of breast cancer treatment in Canada. Eur J Cancer. 2000;36(6):724–35. https://doi.org/10.1016/S0959-8049(99)00340-8.
Hillner BE, Smith TJ. Efficacy and cost effectiveness of adjuvant chemotherapy in women with node-negative breast cancer. A decision-analysis model. N Engl J Med. 1991;324(3):160–8.
Baker MS, Kessler LG, Smuker RC. Site-specific treatment costs for cancer: an analysis of the medicare continuous history sample File., M.H.A.P. Cancer Care and Cost. Ann Arbor, Editor. 1989.
Sculpher M. Subgroups and heterogeneity in cost-effectiveness analysis. Pharmacoeconomics. 2008;26(9):799–806.
Brazier J, et al. Identification, review, and use of health state utilities in cost-effectiveness models: An ISPOR good practices for outcomes research task force report. Value Health. 2019;22(3):267–75.
Papaioannou D, Brazier J, Paisley S, University of Sheffield. NICE DSU Technical Support Document 9: the identification, review and synthesis of health state utility values from the literature. 2010.
Fusco D, et al. An update in breast cancer management for elderly patients. Transl Cancer Res. 2018;7(suppl 3):1–10.
Peasgood T, Ward SE, Brazier J. Health-state utility values in breast cancer. Expert Rev Pharmacoecon Outcomes Res. 2010;10(5):553–66.
Scott NW, et al. EORTC QLQ-C30 reference values manual. 2008.
Guerra RL, et al. Breast cancer quality of life and health-state utility at a brazilian reference public cancer center. Expert Rev Pharmacoecon Outcomes Res. 2020;20(2):185–91.
Crott R, Briggs A. Mapping the QLQ-C30 quality of life cancer questionnaire to EQ-5D patient preferences. Eur J Health Econ. 2010;11(4):427–34.
Arving C, et al. Cost-utility analysis of individual psychosocial support interventions for breast cancer patients in a randomized controlled study. Psychooncology. 2014;23(3):251–8.
Soares MO, et al. Methods to elicit experts’ beliefs over uncertain quantities: application to a cost effectiveness transition model of negative pressure wound therapy for severe pressure ulceration. Stat Med. 2011;30(19):2363–80.
Henson KE, et al. Data resource profile: national cancer registration dataset in England. Int J Epidemiol. 2020;49(1):16–16h.
McGuire A, et al. Effects of age on the detection and management of breast cancer. Cancers (Basel). 2015;7(2):908–29.
Cheung KL, et al. Pathological features of primary breast cancer in the elderly based on needle core biopsies–a large series from a single centre. Crit Rev Oncol Hematol. 2008;67(3):263–7.
Diab SG, Elledge RM, Clark GM. Tumor characteristics and clinical outcome of elderly women with breast cancer. J Natl Cancer Inst. 2000;92(7):550–6.
Gilford DM. The aging population in the twenty-first century: Statistics for health policy. 1988.
Campbell JD, Ramsey SD. The costs of treating breast cancer in the US. Pharmacoeconomics. 2009;27(3):199–209.
Parsekar K, et al. Societal costs of chemotherapy in the UK: an incidence-based cost-of-illness model for early breast cancer. BMJ Open. 2021;11(1):e039412.
Kattlove H, et al. Benefits and costs of screening and treatment for early breast cancer. Development of a basic benefit package. JAMA. 1995;273(2):142–8.
Malin J, Keeler EJR. Los Angeles, Cost-effectiveness and health benefit of breast cancer in women with early breast cancer. 2000.
Stout NK, et al. Retrospective cost-effectiveness analysis of screening mammography. J Natl Cancer Inst. 2006;98(11):774–82.
Ito K, Blinder VS, Elkin EB. Cost effectiveness of fracture prevention in postmenopausal women who receive aromatase inhibitors for early breast cancer. In Presented at the. 2012. Citeseer.
Skedgel C, Rayson D, Younis T. The cost-utility of sequential adjuvant trastuzumab in women with Her2/Neu-positive breast cancer: An analysis based on updated results from the HERA Trial. Value Health. 2009;12(5):641–8.
Younis T, Rayson D, Skedgel C. The cost–utility of adjuvant chemotherapy using docetaxel and cyclophosphamide compared with doxorubicin and cyclophosphamide in breast cancer. Curr Oncol. 2011;18(6):e288.
Younis T, et al. Adjuvant chemotherapy for breast cancer: a cost-utility analysis of FEC-D vs. FEC 100. Breast Cancer Res Treat. 2008;111(2):261–7.
Anderson RN. United States life tables, 1997. Natl Vital Stat Rep. 1999;47(28):1–37.
Drucker A, Skedgel C, Virik K, Rayson D, Sellon M, Younis T. The cost burden of trastuzumab and bevacizumab therapy for solid tumours in Canada. Curr Oncol. 2008;15(3):136–42. https://doi.org/10.3747/co.v15i3.249.
Levy AR, Briggs AH, Demers C, O’Brien BJ. Cost-effectiveness of β-blocker therapy with metoprolol or with carvedilol for treatment of heart failure in Canada. Am Heart J. 2001;142(3):537–43. https://doi.org/10.1067/mhj.2001.116479.
Potvin KR, Younis T, Sellon M, Barnes P, Rayson D. Patterns of trastuzumab use and cost in a single Canadian cancer institute. J Clin Oncol. 2005;23(16_suppl):6070–70. https://doi.org/10.1200/jco.2005.23.16_suppl.6070.
Rao S, Kubisiak J, Gilden D. Cost of illness associated with metastatic breast cancer. Breast Cancer Res Treat. 2004;83(1):25–32.
This study did not receive any research funding. Yubo Wang received the University of Manchester—China Scholarship Council joint scholarship for his PhD study (2018 to 2021).
Ethics approval and consent to participate
Consent for publication
All authors (Yubo Wang, Sean P Gavan, Douglas Steinke, Kwok-Leung Cheung, and Li-Chia Chen) declare no conflict of interest.
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
About this article
Cite this article
Wang, Y., Gavan, S.P., Steinke, D. et al. Systematic review of the evidence sources applied to cost-effectiveness analyses for older women with primary breast cancer. Cost Eff Resour Alloc 20, 9 (2022). https://doi.org/10.1186/s12962-022-00342-7
- Economic evaluation
- Decision-analytic modelling
- Data sources of input parameters
- Older women
- Primary breast cancer