Economic Evaluations of Breast Cancer Care in Low‐ and Middle‐Income Countries: A Scoping Review

Abstract

Background

Understanding the cost of delivering breast cancer (BC) care in low‐ and middle‐income countries (LMICs) is critical to guide effective care delivery strategies. This scoping review summarizes the scope of literature on the costs of BC care in LMICs and characterizes the methodological approaches of these economic evaluations.

Materials and Methods

A systematic literature search was performed in five databases and gray literature up to March 2020. Studies were screened to identify original articles that included a cost outcome for BC diagnosis or treatment in an LMIC. Two independent reviewers assessed articles for eligibility. Data related to study characteristics and methodology were extracted. Study quality was assessed using the Drummond et al. checklist.

Results

Ninety‐one articles across 38 countries were included. The majority (73%) of studies were published between 2013 and 2020. Low‐income countries (2%) and countries in Sub‐Saharan Africa (9%) were grossly underrepresented. The majority of studies (60%) used a health care system perspective. Time horizon was not reported in 30 studies (33%). Of the 33 studies that estimated the cost of multiple steps in the BC care pathway, the majority (73%) were of high quality, but studies varied in their inclusion of nonmedical direct and indirect costs.

Conclusion

There has been substantial growth in the number of BC economic evaluations in LMICs in the past decade, but there remain limited data from low‐income countries, especially those in Sub‐Saharan Africa. BC economic evaluations should be prioritized in these countries. Use of existing frameworks for economic evaluations may help achieve comparable, transparent costing analyses.

Implications for Practice

There has been substantial growth in the number of breast cancer economic evaluations in low‐ and middle‐income countries (LMICs) in the past decade, but there remain limited data from low‐income countries. Breast cancer economic evaluations should be prioritized in low‐income countries and in Sub‐Saharan Africa. Researchers should strive to use and report a costing perspective and time horizon that captures all costs relevant to the study objective, including those such as direct nonmedical and indirect costs. Use of existing frameworks for economic evaluations in LMICs may help achieve comparable, transparent costing analyses in order to guide breast cancer control strategies.

Short abstract

This review summarizes the literature on the cost of breast cancer care in low‐ and middle‐income countries.

Introduction

Breast cancer (BC) is the most commonly diagnosed cancer among women worldwide and the leading cause of cancer death in more than 100 countries [1]. In 2020, there were about 2.3 million new BC cases and about 685,000 BC deaths [1, 2]. These deaths disproportionately occur in low‐ and middle‐income countries (LMICs), where BC mortality rates are rapidly rising [3]. Regions with the highest mortality to incidence ratio include Africa (0.47), South‐Central Asia (0.48), and Melanesia (0.48) [3]. This is in contrast to North America, Australia/New Zealand, and Western Europe, where the BC mortality to incidence ratio is 0.16, 0.17, and 0.18, respectively. Poor outcomes in LMICs reflect the large proportion of women with BC who present with advanced disease and have limited access to diagnosis and treatment [4].

Effective therapeutic options to treat early BC are becoming more widely available at low cost [5, 6, 7]. More widespread deployment of BC diagnostics and treatments could salvage many life years for women in LMICs. Although cancer programs established in low‐resource settings have demonstrated the feasibility of cancer care delivery in LMICs, concerns and misconceptions about affordability of cancer care continue to impede efforts to establish and expand care [8]. Understanding the cost of delivering high‐quality cancer care in LMICs is integral for strategic policy planning and investment in cancer control.

Prior systematic reviews of BC cost in LMICs have largely focused on screening programs and have noted a lack of strong evidence to provide specific recommendations [9, 10, 11]. Reviews that capture the cost of BC diagnosis or treatment in LMICs are limited, as they often consider the cost‐effectiveness of a specific chemotherapeutic or biologic therapy or are largely descriptive [5, 7, 12, 13, 14]. One prior systematic review of the cost of BC care in LMICs, published in 2013, concluded that the evidence base to guide strategies for BC control in LMICs was limited and of poor quality [15]. The majority of economic analyses captured by this review estimated the incremental cost or cost‐effectiveness of a singular diagnostic or therapeutic step, rather than the total cost of the various steps in the breast cancer care pathway. The focus on incremental cost may limit these studies’ applicability given that the breast cancer care pathway is complex and involves multiple diagnostic and therapeutic steps.

Since 2013, many LMICs have made major strides in building capacity to diagnose and treat BC, including the development of cancer centers in low‐resource settings as well as expanded access to inexpensive drugs and novel diagnostic technologies [16, 17, 18, 19]. In 2015, the United Nations Sustainable Development Goals included reducing premature deaths from noncommunicable diseases, of which breast cancer is a considerable part [7]. In order to support governments of LMICs in their commitment to developing and implementing locally appropriate cancer control strategies, researchers have developed frameworks and tools to analyze interventions for effectiveness and affordability in LMICs [7, 20, 21, 22]. Given these strides over the past decade, an updated assessment of the literature is needed to reflect the changing landscape of BC care in LMICs and guide research priorities. This review aims to summarize the scope of literature on the costs of BC care in LMICs, characterize the methodological approaches used in these economic evaluations, and evaluate their methodological rigor.

Materials and Methods

The scoping review was conducted in accordance with the Joanna Briggs Institute methodology for scoping reviews, and the Preferred Reporting Items for Systematic Reviews and Meta‐Analyses‐Scoping Review was referenced to ensure that all suggested reporting items were included [23, 24]. An a priori protocol was used [25].

Search Strategy

Five databases—including MEDLINE (Ovid), Embase (Elsevier), Web of Science (Clarivate Analytics), Global Health (EBSCO), and World Health Organization (WHO) Global Index Medicus—were searched up to March 19, 2020. The search strategy was developed in consultation with a medical librarian and used Medical Subject Headings related to breast neoplasms, costing, and LMICs (supplemental online Appendix 1). Sources of unpublished studies and gray literature—including the Breast Global Health Initiative, Disease Control Priorities 3rd edition, and the World Health Organization—were also searched for relevant articles. The reference lists of articles included in the scoping review were also screened for additional studies.

Eligibility Criteria

Studies included in the review were required to report a cost outcome for BC diagnosis or treatment in LMICs, as defined by 2020 World Bank classifications [26]. We excluded studies that assessed only BC screening, palliative care, or mortality costs as well as studies that did not include any original cost analysis but used previously published cost analyses. Studies that presented aggregate costs for multiple cancers or an entire world region but did not stratify costs for BC or LMICs, respectively, were excluded. Reviews, editorials, and meeting abstracts were also excluded. This scoping review was limited to studies for which manuscripts were available in English.

Screening and Data Extraction

Following the search, all identified records were collated and uploaded to Covidence systematic review software (Veritas Health Innovation, Melbourne, Australia) and duplicates were removed [27]. Studies underwent a primary and secondary screen as described in supplemental online Appendix 2A [28]. Final studies included in the scoping review were divided into two categories: studies that estimated the cost of a singular step in the BC care pathway (e.g., cost of chemotherapy alone, or cost of second‐line treatment for metastatic breast cancer) versus studies that estimated the cost of multiple steps in the BC care pathway (e.g., cost of multiple treatment modalities across breast cancer stages). We created this dichotomy in order to perform a more in‐depth characterization of studies that capture multiple steps in the breast cancer care pathway, as these studies offer an opportunity for more detailed cross‐study comparisons of cost comprehensiveness.

One reviewer (P.E.) extracted data from all studies using a data extraction tool programmed in Research Electronic Data Capture (REDCap) [29]. The data extraction tool was iteratively modified by the research team after pilot data extraction from 5 studies (supplemental online Appendix 2B, 2C) [5, 15, 30]. All studies underwent data extraction for variables related to study characteristics: world region, economic status, study design, BC stage, BC subtype, costing perspective, and time horizon [26]. Studies that estimated the cost of multiple steps in the BC care pathway underwent additional data extraction for variables related to cost estimation, in order to provide a more detailed characterization of cost categories and cost analysis approaches used in these studies. The additional variables related to cost estimation include costing approach, cost categories, cost inputs for each cost category, data sources for cost estimation, cost disaggregation by stage, cost disaggregation by cost input, currency details, cost discounting, inflation adjustments, uncertainty estimation, sensitivity analysis, and quality assessment.

Quality assessment was performed using an established 35‐point economic evaluation checklist by Drummond et al. [31]. Similar to previous reviews of economic evaluation, checklist items that did not apply to any of the reviewed studies were removed [5, 15]. A three‐point response scale was used to grade the quality of each checklist item, including 0 (not considered), 1 (partially considered), 2 (fully considered), and not applicable [32]. The sum of scores for each study was compared with the maximum attainable score for each study. A study was considered to be of high, medium, or poor quality if it scored ≥70%, 51%–69%, or ≤ 50% of its maximum score, respectively [32]. The extracted data were tabulated and summated for all reviewed studies. A second reviewer (C.S.) independently performed data extraction for 15% of studies that estimated the cost of multiple steps in the BC care pathway (n = 5). The percent of agreement score for data extraction was calculated.

Results

Search Results

The search strategy resulted in a total of 6,340 studies: 1,670 from MEDLINE, 1,499 from Embase, 1,570 from Web of Science, 813 from Global Health, and 788 from WHO Global Index Medicus (Fig. 1). After merging the results from all sources and removing duplicates, 3,866 studies remained. After the primary screen of titles and abstracts, 227 studies remained. After a secondary screen with full text review, 91 studies met all eligibility criteria and were included in the scoping review [33, 34, 35, 36, 37, 38, 39, 40, 41, 42, 43, 44, 45, 46, 47, 48, 49, 50, 51, 52, 53, 54, 55, 56, 57, 58, 59, 60, 61, 62, 63, 64, 65, 66, 67, 68, 69, 70, 71, 72, 73, 74, 75, 76, 77, 78, 79, 80, 81, 82, 83, 84, 85, 86, 87, 88, 89, 90, 91, 92, 93, 94, 95, 96, 97, 98, 99, 100, 101, 102, 103, 104, 105, 106, 107, 108, 109, 110, 111, 112, 113, 114, 115, 116, 117, 118, 119, 120, 121, 122, 123]. The full texts of 13/227 articles (6%) were excluded because their full texts were not successfully retrieved. Thirty‐three of 227 (15%) articles were excluded because their English full texts were not available (Fig. 1; supplemental online Appendix 3; supplemental online Table 4). Cohen's Kappa score for the primary and secondary screen were 0.62 and 0.82, respectively. Of the 91 reviewed studies, 58 (64%) estimated the cost of a singular step in the BC care pathway [33, 34, 35, 36, 37, 38, 39, 40, 41, 42, 43, 44, 45, 46, 47, 48, 49, 50, 51, 52, 53, 54, 55, 56, 57, 58, 59, 60, 61, 62, 63, 64, 65, 66, 67, 68, 69, 70, 71, 72, 73, 74, 75, 76, 77, 78, 79, 80, 81, 82, 83, 84, 85, 86, 87, 88, 89, 90] whereas 33 (36%) estimated the cost of multiple steps in the BC care pathway [91, 92, 93, 94, 95, 96, 97, 98, 99, 100, 101, 102, 103, 104, 105, 106, 107, 108, 109, 110, 111, 112, 113, 114, 115, 116, 117, 118, 119, 120, 121, 122, 123]. The percent of agreement for data extraction and the Drummond et al. checklist were 97% and 93%, respectively. Specific characteristics of the studies summarized below are outlined in supplemental online Appendix 3 and supplemental online Tables 1 and 2.

Figure 1.

Preferred Reporting Items for Systematic Reviews and Meta‐Analyses (PRISMA) flow diagram.

Scope of Literature and Study Characteristics

Table 1 outlines the scope of literature on the cost of BC care in LMICs. Of the 91 reviewed studies, 73% were published after 2012 (n = 66). Studies spanned across 38 countries (Fig. 2) [124]. Latin America & Caribbean (n = 26, 29%), East Asia & Pacific (n = 25, 27%), and Middle East & North Africa (n = 19, 21%) had the greatest number of studies, whereas South Asia (n = 8, 9%), Sub‐Saharan Africa (SSA; n = 8, 9%), and Europe & Central Asia (n = 5, 5%) had the fewest number of studies. Specific countries with the highest number of studies were China (n = 16), Iran (n = 14), Brazil (n = 12), and Mexico (n = 8). Of the reviewed studies, 70 (77%) were conducted in upper‐middle‐income countries, 22 (24%) in lower‐middle‐income countries, and 2 (2%) in low‐income countries. Forty‐four studies (48%) were cost analyses or cost of illness studies, whereas 45 studies (49%) were cost‐effectiveness or cost‐utility analyses.

Table 1.

Characteristics of studies in scoping review

Characteristics	Reviewed studies (n = 91)
Publication year (1995–2020)
Before 2013	25 (27)
2013–2020	66 (73)
Number of countries	38
World region
Latin America & Caribbean	26 (29)
Sub‐Saharan Africa	8 (9)
Middle East & North Africa	19 (21)
Europe & Central Asia	5 (5)
South Asia	8 (9)
East Asia & Pacific	25 (27)
Economic status a
Upper‐middle income	70 (77)
Lower‐middle income	22 (24)
Low income	2 (2)
Type of economic evaluation
Cost analysis/Cost of illness	44 (48)
Cost‐effectiveness/Utility analysis	45 (49)
Cost minimization analysis	2 (2)
Study design
Observational	41 (45)
Model‐based	42 (46)
Experimental	2 (2)
Other	6 (7)
Evaluated intervention for cost estimation
Diagnosis	5 (5)
Treatment	55 (60)
Diagnosis and treatment	31 (34)
BC stages
Early	26 (29)
Advanced	15 (16)
All	44 (48)
Unknown	1 (1)
Other (e.g., operable, node+)	5 (5)
BC types
Hormone receptor+ only	13 (14)
HER2+ only	19 (21)
Study perspective a
Health care provider	6 (7)
Health care payer	20 (22)
Health care (not specified)	29 (32)
Patient	11 (12)
Societal	23 (25)
Unknown	10 (11)
Time horizon
<1 year	5 (5)
1–9 years	25 (27)
10–19 years	6 (7)
20–39 years	6 (7)
Lifetime or ≥ 40	19 (21)
Unknown	30 (33)

Data are presented as n (%).

^a Categories are not mutually exclusive, and percentages may sum up to more than 100%.

Abbreviations: BC, breast cancer; HER2, human epidermal growth factor receptor 2.

Figure 2.

Countries represented in reviewed studies.

Five studies (5%) evaluated the cost of diagnosis, 55 studies (60%) evaluated the cost of treatment, and 31 studies (34%) evaluated the cost of both diagnosis and treatment. Forty‐four studies (48%) included all BC stages, whereas 26 studies (29%) and 15 studies (16%) included early or advanced stages, respectively. Thirteen studies (14%) focused on hormone receptor–positive BC, whereas 19 studies (21%) focused on HER2‐positive BC. Fifty‐five studies (60%) used a health care perspective for estimating costs, whereas 23 studies (25%) and 11 studies (12%) used a societal or patient perspective, respectively. Eight studies (9%) presented costing data from two perspectives. The perspectives of 10 studies (11%) were not reported. The time horizons of the studies ranged from 3 months to the lifetime of the study population. Thirty studies (33%) had a time horizon less than 10 years, 12 studies (13%) between 10 and 39 years, and 19 studies (21%) greater than 39 years or lifetime. The time horizons of 30 studies (33%) were not reported (Table 1).

Cost Estimation Characteristics

Table 2 outlines the cost estimation characteristics for the 33 studies that estimated the cost of multiple steps in the BC care pathway. The large majority of these studies (n = 27, 82%) were cost analyses or cost of illness studies (supplemental online Appendix 4). Of these 33 studies, 22 (67%) used a micro‐costing approach (in which the cost of each input was estimated separately to calculate total cost), 5 (15%) used a gross‐costing approach, and 6 (18%) used both. Studies used various data sources to estimate cost and resource use, including information from patients (e.g., interviews, questionnaires, shadowing), medical records, hospital data, government data, insurance data, literature, and expert opinion. The most commonly used sources included government data (n = 16, 48%), hospital data (n = 16, 48%), patients (n = 14, 42%), and medical records (n = 13, 39%; Table 2). Twenty‐three studies (70%) used more than one data source to estimate cost and resource use.

Table 2.

Cost estimation approach and inputs for studies estimating the cost of multiple steps in the breast cancer care pathway

Costing characteristics	Reviewed studies (n = 33) a
Costing approach
Micro	22 (67)
Gross	5 (15)
Micro + gross	6 (18)
Sources for costs and resource use
Patients (e.g., interviews, questionnaires)	14 (42)
Medical records	13 (39)
Hospital finance/administrative data	16 (48)
Government data	16 (48)
Insurance data	4 (12)
Literature	8 (24)
Expert opinion	10 (30)
Cost categories included
Direct medical	33 (100)
Direct nonmedical	9 (27)
Indirect	7 (21)
Inputs into direct medical costs b	33 (100)
Medical visits	25 (76)
Diagnostic studies/pathology	23 (70)
Tumor‐directed medications	30 (91)
Supportive medications	13 (39)
Surgery	27 (82)
Radiotherapy	22 (67)
Hospitalization	26 (79)
Imaging	14 (42)
Laboratory tests and blood services (electrolytes, urinalysis, complete blood count)	22 (67)
Palliative care	10 (30)
Training	4 (12)
Administrative/overhead costs	9 (27)
Unspecified medical costs	4 (12)
Inputs into direct nonmedical costs b , c	9 (100)
Food	7 (78)
Travel/Transportation	9 (100)
Accommodation	5 (56)
Food, transportation, or accommodation for companion	2 (22)
Other (e.g., child tutoring, home help)	1 (11)
Inputs into indirect costs b , c	7 (100)
Lost wages from cancer care	7 (100)
Lost wages from disability/premature mortality	6 (86)
Lost wages of companion	5 (71)

Data are presented as n (%).

^a Data from the 33 studies that estimate the cost of multiple steps in the breast cancer care pathway.

^b Categories are not mutually exclusive, and percentages may sum up to more than 100%.

^c Inputs into direct medical costs, n = 33; inputs into direct nonmedical costs, n = 9; inputs into indirect costs, n = 7.

Studies also varied greatly in the cost categories included in their cost estimation. All 33 (100%) studies included direct medical costs, but only 9 (27%) included direct nonmedical costs and 7 (21%) included indirect costs. Twenty‐three studies (70%) included only one cost category in their cost estimation (e.g., direct medical only), 4 (12%) included two cost categories (e.g., direct medical and direct nonmedical), and 6 (18%) included all three cost categories (Table 2).

The inputs into each cost category varied across studies. For the 33 studies that included direct medical costs, inputs included tumor‐directed medications (n = 30, 91%), surgery (n = 27, 82%), hospitalization (n = 26, 79%), medical visits (n = 25, 76%), diagnostic studies and pathology (n = 23, 70%), radiotherapy (n = 22, 67%), laboratory tests and blood services such as complete blood count, electrolytes, and urinalysis (n = 22, 67%), imaging (n = 14, 42%), and supportive medications (n = 13, 39%). For the nine studies that included direct nonmedical costs, inputs included travel (n = 9, 100%), food (n = 7, 78%), accommodations (n = 5, 56%), and companions’ food, transportation, or accommodation (n = 2, 22%). For the seven studies that included indirect costs, inputs included lost wages from cancer care (n = 7, 100%), lost wages from disability or premature mortality (n = 6, 86%), and lost wages of companion from cancer care (n = 5, 71%; Table 2).

In addition to variations in cost categories and inputs, studies also varied in their cost estimation analysis and reporting (Table 3). Of the studies in which costs were estimated beyond 1 year (n = 23), 13 studies (57%) did not discount costs. Of the studies in which costs were collected from various calendar years (n = 18), 10 studies (56%) did not adjust for inflation. Twenty‐six studies (79%) presented the costs in USD or International Dollars, whereas 7 studies (21%) used other currencies. Seven studies (21%) did not report the currency year. Eighteen studies (55%) did not report any uncertainty with their cost estimation and 24 studies (73%) did not perform a sensitivity analysis. The granularity of the presented cost outcome also varied across studies. Sixteen studies (48%) did not disaggregate costs by BC stage (n = 16). Of the studies that used micro‐costing (n = 28), 6 (21%) did not disaggregate costs by cost inputs (Table 3).

Table 3.

Cost estimation reporting and analysis for studies estimating the cost of multiple steps in the breast cancer care pathway

Cost reporting and analysis characteristics	Reviewed studies (n = 33)a
Cost discounting b
Yes	10 (43)
No	13 (57)
Not applicable	10
Inflation adjustment b
Yes	8 (44)
No	10 (56)
Not applicable	15
Currency
USD or International Dollars	26 (79)
Other currencies	7 (21)
Currency year reported
Yes	26 (79)
No	7 (21)
Cost estimation uncertainty reported
Yes	15 (45)
No	18 (55)
Sensitivity analysis performed
Yes	9 (27)
No	24 (73)
Cost disaggregation by BC stage
Yes	17 (52)
No	16 (48)
Costs disaggregation by inputs b
Yes	22 (79)
No	6 (21)
Not applicable	5

Data are presented as n (%).

^a Data from the 33 studies that estimate the cost of multiple steps in the breast cancer care pathway.

^b Percentage scores do not include studies for which the variable was not applicable. Cost discounting was applicable to studies in which costs were estimated beyond 1 year; inflation adjustment was applicable to studies in which costs were collected from various calendar years; cost disaggregation by inputs was applicable to studies in which a micro‐costing approach was used.

Abbreviation: BC, breast cancer.

Quality Assessment

Table 4 outlines the quality assessment of the 33 studies that estimated the costs of multiple steps in the BC care pathway using the economic evaluation checklist by Drummond et al. [31]. Of the 33 studies, 24 (73%) scored ≥70% of the maximum possible score (high quality) whereas 9 (27%) scored 51%–69% of the maximum possible score (medium quality) [32]. In both the “Cost and Effect Estimation” and “Analysis and Interpretation” checklist categories, five studies (15%) scored ≤50% of the maximum score (poor quality). Total and disaggregated scoring for each study is outlined in supplemental online Appendix 3 and supplemental online Table 3.

Table 4.

Quality assessment of breast cancer costing studies

Percentage of maximum score	Study design	Cost + effect estimation	Analysis + interpretation	Total
≤50%	0 (0)	5 (15)	5 (15)	0 (0)
50%–70%	0 (0)	7 (21)	12 (36)	9 (27)
≥70%	33 (100)	21 (64)	16 (48)	24 (73)

Data are presented as n (%).

Data from the 33 studies that estimate the cost of multiple steps in the breast cancer care pathway.

Quality assessment based on Drummond checklist used; studies scored based on applicable categories [30].

Discussion

This scoping review of literature on the cost of BC care in LMICs yielded 91 articles across 38 countries, with the majority of economic evaluations representing upper‐middle‐income countries in Latin America & Caribbean, East Asia & Pacific, and Middle East & North Africa. The most commonly used costing perspective was that of the health care system. The time horizon for costs ranged from 3 months to lifetime; however, about one third of studies did not report a time horizon. Of the studies that estimated the cost of multiple steps in the BC care pathway, the majority used a micro‐costing approach. All of these studies included direct medical costs, whereas a minority included direct nonmedical or indirect costs. Although the majority of these studies were of high quality, several studies lacked several important costing details.

Cost of BC care in LMICs remains an understudied area. A 2013 systematic review of BC control economic analyses in LMICs (including BC screening, diagnosis, and treatment) by Zelle et al. identified 24 studies and reported limited economic evidence on BC control [15]. Although our scoping review had a narrower focus (excluding studies only focused on BC screening), we identified 25 studies published before 2013, likely due to searching a greater number of databases. Since 2013, there has been a substantial increase in the number of BC costing studies in LMICs, suggesting a growing interest in BC costing data in low‐resource settings.

Despite the increasing number of studies, the majority of data have been largely limited to upper‐middle‐income countries. Low‐income countries make up 21% of LMICs (29/135) but represented only two (2%) of the reviewed studies [26]. One of the two studies in low‐income countries focused on out‐of‐pocket expenses incurred by patients obtaining free BC care in Haiti [106]. The other included minimal original data and relied heavily on costing studies from a middle‐income country (Bolivia) to evaluate the cost‐effectiveness of trastuzumab in SSA [81]. Specific world regions were also underrepresented. There were few studies from SSA—only eight (9%) from a region representing 34% of all LMICs (46/135). Similarly, there were only five studies (5%) from Europe & Central Asia, a region that represents 15% of all LMICs (20/135) [26]. These findings are consistent with a recent systematic review of BC treatments costs by stage worldwide by Sun et al., which only included five studies from LMICs, none of which were from low‐income countries or countries in SSA [5]. The underrepresentation of low‐income countries and countries in SSA may represent nascent cancer programs in these countries or limited research capacity to conduct such economic evaluations. However, BC costing data may be especially useful in these areas, as many of these countries are expanding national cancer care programs and making difficult decisions on resource allocation. Future costing studies for BC control should be prioritized in Sub‐Saharan Africa as well as in low‐income countries with developing BC programs.

One of the challenges in conducting costing studies in LMICs is the paucity of commonly accepted guidelines to design, conduct, and report economic evaluations. Several upper‐middle‐income countries (e.g., China, Brazil, Colombia) and lower‐middle‐income countries (e.g., Egypt, Bhutan, Philippines) have country‐specific guidelines [125]. Others use frameworks that have been designed for use in LMICs—such as the International Decision Supportive Initiative (iDSI) Reference Case or the WHO guide to cost‐effective analysis (WHO‐CHOICE) [21, 22]. The development of a unified framework, which can be adapted across different settings, may help promote transparency, consistency, and comparison.

The studies included in this scoping review showed mixed adherence to existing frameworks as it relates to transparency of cost perspective and time horizon. For example, 11% of studies did not report their costing perspectives, which limits the interpretability and utility of the presented data. Although a study's perspective depends on its purpose and audience, the current costing literature for BC care in LMICs is biased toward costs incurred by the health care sector—60% of reviewed studies used a health care perspective. A study's time horizon also depends on its purpose; for example, studies that estimate the cost of BC diagnosis may require a substantially shorter time horizon than those that estimate treatment costs [21, 22, 126]. However, in the reviewed literature, 33% of studies did not report their time horizon at all. One of the challenges of conducting costing studies with longer time horizons is the limited availability of empirical cost data in LMICs. For example, when studies in this review relied on patient interviews for costing data, they had particularly short time horizons, spanning 6 months to 2 years (likely due to limitations in patient recall of cost estimates) [94, 113, 118]. However, as previously described, time horizons do not necessarily need to be limited by the availability of empirical data, as economic evaluations may use validated imputation methods for missing data [22]. Researchers should strive to use and report a costing perspective and time horizon that comprehensively captures all costs relevant to the study objective.

For the 33 studies that estimated the cost of multiple steps in the BC care pathway, we examined the applied costing methodologies and included cost categories. The large majority of studies used a micro‐costing approach, which tends to be a more comprehensive form of costing that is less likely to underestimate costs compared with gross‐costing methods [127]. However, only 21% of studies explicitly reported the quantities of resources separately from their unit costs, suggesting suboptimal use or reporting of micro‐costing [68, 95, 98, 101, 103, 105, 112]. Micro‐costing may be very time intensive and onerous in low‐resource settings, as it requires disaggregation by input. Gross‐costing may provide some benefits such as ease of estimating overhead/administrative or training costs, which were captured by only a minority of reviewed studies [128]. Future BC costing studies in LMICs may consider using both micro‐ and gross‐ costing methods for different cost measures.

Furthermore, analysis of cost inputs for these 33 studies revealed that BC economic evaluations were often missing key costs categories. The large majority of studies were missing direct nonmedical and indirect costs. Exclusion of these costs inputs likely biases cost outcomes toward costs incurred by the health care sector, while missing potentially substantial costs incurred by patients or society [129]. Use of a societal or patient perspective may encourage the inclusion of direct nonmedical and indirect costs in future economic evaluations.

The overall quality of the 33 studies that estimated the cost of multiple steps in the BC care pathway ranged from medium to high quality based on the Drummond et al. checklist. In the 2013 Zelle et al. review of BC economic evaluations in LMICs (which included studies on BC screening), 11 out of the 24 studies had poor quality. In this scoping review, the three highest‐quality studies were all cost‐effectiveness studies published after 2013 and explicitly used the WHO‐CHOICE methodology (Zelle et al., Zelle et al., and Niens et al.) [98, 101, 103]. In contrast, the two lowest‐quality studies were the earliest studies in this review (Arredondo et al., Yazihan et al.) [91, 92]. Overall, the most commonly missed Drummond et al. checklist items included inclusion of time horizon, costing perspective, discounting, inflation adjustment, uncertainty estimation, consideration of productivity changes, and separate reporting of resource quantities and unit costs. The lack of uncertainty estimation (55% of studies) and sensitivity analyses (73% of studies) is especially notable. Given that costs vary within local settings and all costing analyses require some level of estimation, sensitivity analyses and uncertainty estimates are essential to strengthen confidence in the accuracy and generalizability of the results. Use of existing frameworks, such as WHO‐CHOICE or the iDSI Reference Case, may promote more consistent and transparent study designs, analyses, and reporting of future BC economic evaluations in LMICs [21, 22].

This scoping review has several strengths. The review utilizes a systematic search strategy developed in consultation with a medical librarian that spans across five databases and gray literature. In addition, the review takes an exhaustive approach with evaluating the methodology of the economic evaluations, detailing the included cost categories and the comprehensiveness of inputs included in these categories. Nonetheless, this review also has some limitations. First, the review excluded studies with non‐English full texts, which may have introduced language bias as 33 articles were excluded because of language restrictions. Despite these exclusions, the review included studies from all relevant regions, resulting in more studies than any of the prior reviews. Second, only one reviewer performed data extraction from all reviewed studies. An independent, second reviewer performed data extraction for a fraction of studies and the percent of agreement for data extraction and the Drummond et al. checklist were 97% and 93%, respectively. Third, our quality assessment of reviewed articles was based on a checklist that grants the highest scores for full reporting of all domains. However, several checklist items were not applicable for many studies, and thus, each remaining checked item carried a disproportionate weight. Therefore, the quality scores for these studies should be interpreted with caution. Other assessment tools for economic evaluations share similar limitations [130].

Conclusion

This scoping review highlights a substantial increase in the number of BC economic evaluations in LMICs in the past decade. Despite this growing body of literature, there remain limited data from low‐income countries, especially those in Sub‐Saharan Africa. Future BC economic evaluations should be prioritized in these countries. The current literature was biased toward costs incurred by the health care sector, and as such, direct nonmedical costs and indirect costs were often not included. Although most studies assessing multiple steps in the BC pathway were high quality, notable gaps included missing specification of time horizon, cost estimation uncertainty, and sensitivity analyses. Researchers should strive to use and report a costing perspective and time horizon that captures all costs relevant to the study objective. Use of existing frameworks for economic evaluations in LMICs (such as WHO‐CHOICE or the iDSI Reference Case) may help achieve transparent, comparable costing analyses that can be used to guide BC control strategies.

Author Contributions

Conception/design: Parsa Erfani, Kayleigh Bhangdia, Catherine Stauber, Jean Claude Mugunga, Lydia E. Pace, Temidayo Fadelu

Provision of study material or patients: Parsa Erfani, Kayleigh Bhangdia, Temidayo Fadelu

Collection and/or assembly of data: Parsa Erfani, Kayleigh Bhangdia, Catherine Stauber

Data analysis and interpretation: Parsa Erfani, Kayleigh Bhangdia, Catherine Stauber, Jean Claude Mugunga, Lydia E. Pace, Temidayo Fadelu

Manuscript writing: Parsa Erfani, Kayleigh Bhangdia, Catherine Stauber, Jean Claude Mugunga, Lydia E. Pace, Temidayo Fadelu

Final approval of manuscript: Parsa Erfani, Kayleigh Bhangdia, Catherine Stauber, Jean Claude Mugunga, Lydia E. Pace, Temidayo Fadelu

Disclosures

The authors indicated no financial relationships.

Supporting information

See http://www.TheOncologist.com for supplemental material available online.

Appendix S1. Supporting Information.

Appendix S2. Tables.