Introduction

Worldwide, breast cancer is the most common cancer in women [1]. Although breast cancer-related lymphoedema (BCRL) is not the most prevalent complication after treatment for breast cancer [2], it is internationally recognized as one of the most dreaded morbidities. Since the introduction of more effective treatment modalities [3,4,5,6] increases the number of breast cancer survivors, the amount of patients dealing with long-term side effects, such as lymphoedema, rises likewise [7]. BCRL is caused by a decreased lymphatic transport capacity and/or increased lymphatic load after which fluid accumulates in the extracellular spaces of soft tissues, resulting in swelling [8]. Today, pooled data reveals a BCRL incidence rate of 16.6% [9].

Besides an impact on functional and psychosocial well-being [10], there can be an additional deleterious effect of lymphoedema on patients in terms of financial costs [11, 12]. Daily living can be affected by copayments for the increase in medical and therapeutic consultations, as well as by other direct costs for compression garments and other (in)direct therapy-related expenses [11]. Moreover, financial burdensome can be emphasized through the impact of (advanced) lymphoedema on career and employment [12]. This happens for instance when a transition from fulltime to part-time employment is required in order to spend more time on complex care [12]. Besides the lymphoedema which requires appropriate treatment, complications secondary to BCRL, such as repeated infections, may arise as well [13]. These episodes need early antibiotic therapy and may require hospitalization, increasing the costs of care even more [14].

According to the recommendations of the International Society of Lymphology (ISL), BCRL needs to be treated with decongestive lymphatic therapy (DLT) [15]. This is a two-stage treatment program, consisting of different conservative treatment modalities. During the first or intensive phase, lymphoedema is maximally reduced. This phase consists of skin care, manual lymph drainage (MLD), multi-layer bandaging and exercise therapy. The second or maintenance phase aims to conserve and optimize the results obtained in the first phase. It consists of skin care, compression by a low-stretch compression sleeve, exercises and MLD [16]. Although DLT is recognized as the gold standard for conservative treatment of lymphoedema [15, 17], reimbursement for DLT has been hampered by a lack of rigorous research evidence [8]. Additionally, current literature on the financial burden of BCRL treatment is extremely limited. An overview between patient-borne and society-borne costs within this financial burden is missing. However, this is essential to estimate the actual economic impact of BCRL for patients as for society.

Therefore, the aim of this review was to make an overview of the currently available literature on direct and indirect patient-borne as well as society-borne costs associated with the treatment of BCRL and its sequelae.

Methods

Literature search and inclusion criteria

According to the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) guideline [18] (www.prisma-statement.org), a systematic review of the literature was performed. This review has been registered on PROSPERO (https://www.crd.york.ac.uk/PROSPERO) with registration number CRD42018114649. In order to identify eligible studies, four electronic databases were screened on October 1, 2018: PubMed, Web of Science, EMBASE and Cochrane Clinical Trials. A PICOS search strategy was built up, resulting in a Boolean search where following indexing terms (i.e. MeSH for Pubmed and Cochrane, Emtree for EMBASE) and keywords were combined: ‘breast cancer(P)’, ‘lymphoedema(P)’, ‘decongestive lymphatic therapy(I)’, ‘treatment(I)’, ‘economic analysis(O)’, ‘economic evaluation(O)’ and ‘costs(O)’. A comparison was not defined (not applicable). Equivalent searches were executed in all four databases, although modifications in keywords were included due to differences in usage of indexing terms. When using Web of Science, an additional restriction was added to the search with the filter ‘document type: Article’, and in EMBASE, the search was limited to ‘Articles’ or ‘Articles in press’ and studies based on ‘Humans’. In Appendix, an overview of the applied search strategies for the different databases is presented.

The screening for eligible articles was twofold and performed by two raters (T.D.V. and N.G.). The first screening upon title and abstract was achieved for all references in each database, in order to assess which articles were relevant for further scrutiny. Thereafter, a second screening on the full texts of the selected articles was performed. Both screenings were based upon predetermined inclusion and exclusion criteria, reported in Table 1. In case of disagreement between the reviewers regarding the in- or exclusion of studies, consensus was reached during a meeting.

Table 1 Eligibility criteria used in both screenings
Full size table

Methodological quality assessment and data extraction

To assess the methodological quality of the selected full texts, the 19-item NICE checklist for (partial) economic evaluations provided by the National Institute for Health and Care Excellence (NICE) [19] (https://www.nice.org.uk/process/pmg20/chapter/incorporating-economic-evaluation) was used. Full texts were evaluated by both reviewers (T.D.V. and N.G.). As the NICE checklist initially is designed for the UK, some minor adjustments in questions were necessary to generalize the feasibility of the questions to all countries [19]. An item was scored ‘1’ if adequate information was provided and bias was unlikely. An item was scored ‘0’ if the criterion was not met. An item was scored ‘?’ if the required information was lacking. Afterwards, the total methodological quality was expressed as the sum of all items receiving score ‘1’ (Table 2). In case disagreement occurred between reviewers regarding assigning a score to an item, consensus was sought during a meeting. Additionally, according to the Dutch Cochrane Centre guidelines, levels of evidence were determined for all selected studies (http://netherlands.cochrane.org).

Table 2 Overview of the methodological quality of the eight included studies (NICE checklist)
Full size table

Data on study design, research question, study region, number of participants, inclusion and exclusion criteria, timespan, applied treatment for BCRL, cost- (and other) related outcome measures and cost-related main results were extracted and summarized from the included full texts in Table 3. If studies reported both quantitative and qualitative data concerning the economic burden of BCRL, only quantitative data was extracted. If studies compared treatment costs for patients with and without BCRL, or compared (so-called) standard treatment costs and an experimental/model-based treatment cost, only the BCRL treatment costs and standard treatment costs were mentioned. To increase the interpretability of the amount of costs in the different currencies, we converted the costs that are reported in Euros, Australian Dollars or British Pounds in the ‘Results’ and/or ‘Discussion’ sections, into the USD$ currency and added them in parentheses next to the original currency. This currency exchange is based on the actual exchange rate at the time of the online publication of the article (month, year): €1=USD$1.41 (August 2009) [20] and USD$1.18 (October 2017) [21]; 1A$=USD$0.85 (December 2014) [22] and USD$0.76 (August 2016) [12]; £1=USD$1.43 (February 2018) [23].

Table 3 Table of evidence with characteristics of the eight included studies
Full size table

Results

Study selection

At first, the search yielded 387 references, including duplicates. After the first screening upon title and abstract for each selected database, 28 full texts were retrieved for further scrutiny. After a second screening upon eligibility criteria (Table 1), duplicates (n = 14) were removed. Finally, eight studies were included for the ‘Results’ section of this review: 4 cohort studies [14, 22, 24, 25] and 4 cross-sectional studies [12, 20, 26, 27] respectively. Figure 1 provides a detailed flowchart of the search strategy and selection procedure.

Fig. 1
figure 1

Flowchart of the Boolean search and selection procedure (PRISMA)

Full size image

Methodological quality

An overview of the risk of bias and level of evidence of the included studies is presented in Table 2. Regarding study quality, scores for the (partly) economic evaluations in both cohort and cross-sectional studies ranged between 6/11 and 8/11. A question that frequently scored negative or of which information was lacking, was the following: ‘Are all important parameters whose values are uncertain subjected to appropriate sensitivity analysis?’, because in most cases, the aim of the studies was to provide an overview of costs, rather than to make a cost-effectiveness evaluation. According to the Dutch Cochrane Centre guidelines, levels of evidence ranged between A2 [14, 22, 25] and B [12, 20, 24, 26, 27].

Characteristics of the included studies

Altogether, costs were analyzed of 2421 patients with BCRL from 6 out of 8 included studies which were all women [12, 14, 22, 24, 25, 27]. Two studies did not report the amount of patients upon which their cost-related outcomes were based, nor did they specify the gender of the included patients [20, 26]. Mean age of the included patients ranged between 49 [14] and 63 years [25]. One study did not define mean age [12], and in one study, this was not mentioned since results were based on a hypothetical decision model [27]. Study regions comprised the USA [14, 24,25,26,27], Australia [12, 22] and Finland [20].

Costs related to BCRL

The timespan in which costs were estimated in the different studies ranged between 12 months [20, 22, 25, 27] and 24 months [14, 24]. Either, these costs were retrieved from a convenience sample of patients with BCRL during an arbitrary follow-up of 12 months [20, 25], during the first 12 months after surgery for breast cancer [27], between 6 and 18 months after surgery [22] or during the first 24 months after surgery [14, 24].

Three studies [12, 22, 25] investigated patient-borne costs related to BCRL. Of these, two studies made a distinction between direct (i.e. costs directly related to the treatment for BCRL such as costs for therapeutic measures, physician fees, drugs, compression therapy/garment) and indirect (i.e. productivity losses; values of lost income, unpaid help and lost unpaid work) patient-borne costs [22, 25]. Mean direct costs per patient per year ranged between USD$2306 [25] and USD$2574 [22]. Indirect costs ranged between USD$3325 [25] and USD $5545 [22] costs per year. In the article by Boyages et al., the overall mean patient-borne costs for BCRL per patient per year were provided, resulting in an average of A$977 (= USD$742.52) per year [12]. Hereby, no distinction between direct and indirect costs was made.

The five remaining studies [14, 20, 24, 26, 27] discussed medical costs from a societal perspective. These included costs collected from claims data from (national) insurers [14, 20], physician Medicare fees [26, 27], hospitalization charges [24, 27] and/or manufacturer’s and service providers’ prices [20]. In these studies, no separate overview of out-of-pocket costs borne by patients was provided. One study showed that the average of non-cancer-related medical costs for BCRL was estimated on USD$45,896 per patient during 2 years (USD$22,948 per patient per year), of which USD$1083 per patient was charged for physical therapy and supplies [14]. In Bilir et al., the total 1-year economic impact with direct and indirect costs was USD$1,984,529 for standard assessment and lymphoedema treatment in 627 patients (USD$3165 per patient per year) [27]. Three studies provided an overview of solely direct costs [20, 24, 26]. Direct BCRL-related healthcare charges due to hospitalization (e.g. for systemic infections) were estimated on USD$58,088 per patient during 2 years (USD$29,044 per patient per year) [24]. Direct treatment costs after 1 year of DLT per patient were estimated on USD$3125 [26]. In Finland, total direct costs per patient treated with DLT were €799 (= USD$1126.60) per year [20]. An overview of the extracted data is shown in Table 3.

Discussion

The purpose of this systematic review was to provide an overview of the direct and indirect patient-borne as well as society-borne costs associated with the treatment of BCRL and its sequelae (Table 3 ‘main findings: costs’).

Three out of eight of the included studies were prospective cohort studies with sufficient sample size and follow-up. These studies were graded with a level of evidence A2 [14, 22, 25]. However, scores on methodological quality in terms of risk of bias of the included studies were relatively similar to each other.

This review reveals that BCRL imposes a substantial economic burden on patients and society. When solely direct costs are taken into account, in most cases, a significant proportion of costs is spent on physical therapy sessions and materials (e.g. compression garment), medication and hospital admissions in case of infections. During a 2-year post-operative period, patients with BCRL required significantly more hospitalizations and nearly seven times higher healthcare charge per patient compared to  patients without BCRL (USD$141,388 vs. USD$21,141 per patient, respectively) [24]. If productivity losses were taken into account as well, the financial burden increased even more.

In the article by Stout et al., direct treatment costs associated with a traditional model of DLT were compared with costs associated with a prospective surveillance model [26]. In the USA, the cost to manage early-stage BCRL per patient per year using a prospective surveillance model was USD$636. In contrast, the costs associated with DLT using the traditional model was USD$3125 [26], highlighting the importance of an early treatment onset in favor of less invasive treatment expenses due to fewer treatment sessions and less material required.

This review comprised only one study that investigated the treatment cost for DLT in a European country, whereby results showed an average cost of €799 (= USD$1126.60) per patient per year [20]. However, more information is available concerning treatment costs for lower limb lymphoedema in European settings. Recently, Gutknecht et al. performed in Germany an observational cross-sectional study in patients with chronic lymphoedema or lipolymphoedema in order to analyze all the direct and indirect costs for patients, health insurance and society [21]. The average total cost for each patient per year was €5784 (= USD$6825.12), of which €4445 (= USD$5245.1) (76%) were direct costs and €1338 (= USD$1578.84) (24%) were indirect costs. Patient-borne costs were €648 (= USD$764.64) on average per year, wherein the highest costs were for MLD and disability costs (e.g. prescription fees including private costs for remedies and aids, payments for physician visits, hospitalization and rehabilitation, skin care products) [21]. Each year, a mean cost of €2510 (= USD$2961.80) per patient is spent on MLD, which was considered the main cost factor for statutory health insurances [21]. However, as this study relies on costs regarding the treatment for lower limb oedema, it was not included in our review. Likewise, in another recently published study of Moffatt et al., the aim was to develop and evaluate health service and patient outcomes using an appropriate model of care within a London-based primary care trust [23]. Patients with chronic swelling of the arm(s) or leg(s) were recruited and treated for a period of 6 months. Results of this study showed the benefits of a service model for chronic oedema, with clinical improvements due to a reduction in limb volume and reduced complications. Recourses moved from the acute care setting to lower-cost interventions in community: overall costs reduced from £50,171 (= USD$71,744.53) before implementation, to £27,352 = USD$39,113.96) within the first 6 months and subsequently £17,618 (= USD$25,193.74) between 6 months and 1 year [23].

Several limitations of the included studies of this review need to be discussed. First and foremost, studies investigating the financial costs related to BCRL by making use of claims data [14, 24, 27] are likely to underestimate the real cost rates [28]. Because claims data are designed for billing purposes, they only offer information of patients who are insured. Thus, they only provide an estimation of the costs related to BCRL as they do not yield information about patients with BCRL without health insurance [29]. Furthermore, one should notice that, in case only direct costs related to hospitalizations are taken into account [24], an important underestimation of the complete (direct) costs of BCRL occurs. Evaluation of resource utilization and charges associated with outpatient care would provide a more complete assessment of the impact related to BCRL [24].

Difficulties are being experienced regarding the comparability, transferability and generalizability of the present study results. Transferability is defined as the extent to which the results of a study hold true for a different population or setting [19, 30]. Since different continents, even different states/countries within the same country/continent, are subjected to different healthcare insurance policies and reimbursement procedures, it is difficult to transfer the amount of healthcare costs derived in the USA [14, 24,25,26,27] or Australia [12, 22] to European countries and vice-versa. Besides that, differences in money currencies between countries make the amount of costs derived in the different studies hard to compare. Generalizability is defined as the extent to which the results of a study can be generalized to the population from which the sample size was drawn [19, 30]. As stated by Dean et al., even findings derived from studies conducted solely in the USA are difficult to compare over time, since some of these investigations [14] conducted in the past are predate the 2010 Affordable Care Act that expanded coverage for cancer-related care [25]. Another example is the following: in Shih et al., the study sample was limited to working-age women (mean age 48.8 years); therefore, their findings regarding medical costs may not be generalizable to elderly with BCRL [14].

Limitations and strengths

In this review, literature searches were limited to mainly (bio)medical databases. The NHS Economic Evaluation Database (NHS EED) focusses primarily on the economic evaluation of healthcare interventions [31]. As a result, combining databases such as PubMed and NHS EED should have been an optimal search strategy for economic evaluations [31, 32]. Therefore, a post hoc search was performed on the NHS EED database on October 19, 2018 (https://www.crd.york.ac.uk/CRDWeb). However, this search yielded no additional eligible records.

The present systematic review contains several strengths. Firstly, it has compliance with the PRISMA guideline [18]. Furthermore, to our knowledge, this is the first overview of reported direct and indirect patient-borne as well as society-borne costs specifically associated with the treatment of BCRL, in literature. Lastly, the screening and data extraction process was performed by two blinded researchers.

Knowledge of costs related to BCRL not only improves the understanding of the economic burden of this morbidity but also launches a baseline of comparison for future cost-analytic or cost-effectiveness studies [14]. Therefore, future studies on the effectiveness of treatment modalities for BCRL should consider defining health economic analyses a priori in order to be able to withdraw proper high-quality conclusions based on cost-effectiveness outcomes such as the Incremental Cost-Effectiveness Ratio (ICER) and/or quality-adjusted life-years (QUALI). An appropriate time-horizon (≥ 12 months) should be defined and both incremental (direct and indirect) cost elements from a patient and societal perspective should be considered and collected prospectively. Additionally, it is recommended to include a generic health-related quality of life questionnaires such as the EQ-5D-5L and utility instrument to allow comparisons across interventions and populations.

Conclusion

This review reveals that BCRL imposes a substantial economic burden on patients and society. In the USA, patient-borne direct costs related to BCRL range between USD$2306 and USD$2574 per patient per year. Patient-borne indirect costs range between USD$3325 and USD$5545 per patient per year. Mean direct treatment costs after 1 year of DLT ranged between €799 (= USD$1126.60) and USD$3165. However, these conclusions are based on limited research data and due to the differences in (public) insurance protocols and currencies, it is difficult to compare costs between countries. Therefore, more standardized high-quality health economic analyses among this field are required. Additionally, recent economic analyses related to BCRL treatment in Europe, Asia, Africa and South America are lacking. Worldwide, further scrutiny of the economic impact of DLT for BCRL in clinical settings is needed.