What Are the Real Procedural Costs of Bariatric Surgery? A Systematic Literature Review of Published Cost Analyses

This review aims to evaluate the current literature on the procedural costs of bariatric surgery for the treatment of severe obesity. Using a published framework for the conduct of micro-costing studies for surgical interventions, existing cost estimates from the literature are assessed for their accuracy, reliability and comprehensiveness based on their consideration of seven ‘important’ cost components. MEDLINE, PubMed, key journals and reference lists of included studies were searched up to January 2017. Eligible studies had to report per-case, total procedural costs for any type of bariatric surgery broken down into two or more individual cost components. A total of 998 citations were screened, of which 13 studies were included for analysis. Included studies were mainly conducted from a US hospital perspective, assessed either gastric bypass or adjustable gastric banding procedures and considered a range of different cost components. The mean total procedural costs for all included studies was US$14,389 (range, US$7423 to US$33,541). No study considered all of the recommended ‘important’ cost components and estimation methods were poorly reported. The accuracy, reliability and comprehensiveness of the existing cost estimates are, therefore, questionable. There is a need for a comparative cost analysis of the different approaches to bariatric surgery, with the most appropriate costing approach identified to be micro-costing methods. Such an analysis will not only be useful in estimating the relative cost-effectiveness of different surgeries but will also ensure appropriate reimbursement and budgeting by healthcare payers to ensure barriers to access this effective treatment by severely obese patients are minimised.


Introduction
Bariatric surgery is considered the most efficacious treatment for severe and complex obesity [1] and can be performed using a number of different surgical techniques. In 2013, Roux-en-Y gastric bypass (RYGB) was the most common procedure, followed by sleeve gastrectomy (SG) and adjustable gastric banding (AGB), accounting for 45, 37 and 10% of surgeries performed respectively worldwide [2]. Recently, however, SG has surpassed RYGB as the most common procedure in some jurisdictions, with SG accounting for over 50% of bariatric surgeries performed in the USA [3].
Evidence from retrospective studies suggest that RYGB and SG are likely to result in the greatest amount of weight loss and improvement of comorbidities, but have greater risks and less flexibility than AGB, which is associated with less reliable long-term weight loss [4]. However, evidence from randomised controlled trials The original version of this article was revised because of an error in the author group.
comparing the different surgeries is limited [5,6], as is information on the relative cost-effectiveness of the three surgery types [7]. The lack of the latter evidence could be due partly to a dearth of cost information for the three procedures. Detailed cost estimates (i.e. expenditures involved in acquiring resources that are necessary for the delivery of care) are also a requisite for the development of appropriate reimbursement rates by healthcare payers [8]. For example, the National Health Service (NHS) in the UK has two main tariffs used to reimburse physicians/ hospitals for performing bariatric surgery [£4028 (US$5771) for AGB and SG and £4608 (US$6602) for RYGB] [9]. However, the underlying resources and costs attributed to these tariffs are unclear, which could lead to either over-or underpayment of providers of bariatric surgery.
To fill these evidence gaps, a multi-centre randomised controlled trial, known as the By-Band-Sleeve (BBS) study, is being conducted comparing both the relative effectiveness and cost-effectiveness of RYGB, AGB and SG [5,10]. In comparison to other bariatric surgery trials, the BBS study will assess both clinical and economic outcomes for the three most common approaches to bariatric surgery, in the largest sample size studied in a comparative trial to date (expected to randomise 447 patients per group), over a substantial followup period (36 months). However, an important first step in estimating the economic outcomes in the BBS study will be to obtain detailed and 'accurate' costs of the three types of bariatric surgery.
The costs associated with bariatric surgery are a function of the resources consumed and the unit costs associated with those resources. These parameters can be identified using a number of different approaches, but can broadly be separated into methods used to identify cost components (gross costing or micro-costing) [11] and methods used to value cost components (top-down costing and bottom-up costing) (see Box 1 for definitions) [12]. The choice of method is usually dependent on the context, data and funding available to conduct a costing study, but it is important to note that the application of different methods may result in variations in the magnitude and therefore the accuracy and reliability of the estimated costs [13,14]. Detailed cost estimates, derived using transparent methods, are not only important to provide appropriate reimbursement but also act as accurate inputs in analyses comparing the relative cost-effectiveness of the different surgical approaches and will allow healthcare providers to budget judiciously, potentially minimising barriers to access surgery for the increasing number of severely obese patients [15]. Therefore, our study aimed to review the literature and identify robust estimates of the procedural costs of the main types of bariatric surgery. More specifically, we will summarise the international literature on procedural bariatric surgery costs and assess their comprehensiveness, accuracy and reliability.

Eligibility Criteria
Published costing studies in the English language reporting detailed, per-case (i.e. per-patient) cost estimates associated with performing any type of bariatric surgery were considered. To be included, a cost analysis had to report two or more cost components related to performing the procedure. For example, a study reporting a breakdown of the total cost of surgery in terms of inpatient stay, personnel and equipment costs would be included in the review. Studies only reporting the total cost of surgery with no breakdown into individual cost components (i.e. an aggregated cost) were excluded because it is impossible to understand from total costs what underlying resources were included. Studies only reporting detailed total costs for a single component (e.g. equipment costs) were also excluded, even if they broke down the single component into sub-components (e.g. total equipment costs broken down into maintenance costs, operating costs and personnel costs for cleaning). Studies only assessing the cost of healthcare utilisation either before or after bariatric surgery were also excluded.

Information Sources, Search and Data Collection
Two databases, Ovid MEDLINE and PubMed, were searched from inception to January 16, 2017 (Appendix). Search terms were initially developed for three different categories, costing terminology, types of bariatric surgery and obesity nomenclature, but subsequently refined to increase the sensitivity of the searches. No restrictions were initially placed on the language of the articles to ensure a large number of relevant studies Box 1 Definitions of methods for identifying and valuing cost components Methods for identifying cost components Gross-costing: Involves identifying cost components at a highly aggregated level (e.g. costing an intervention based only on the associated inpatient days) [11]. Micro-costing: A precise method, where an attempt is made to identify every input consumed in the treatment of a particular patient [11].

Methods for valuing cost components
Top-down costing: An approach where relative value units such as hospital days or some other metric are used to separate out relevant costs from comprehensive sources (e.g. the finance department's annual accounts) and apportion them to individual services or procedures [12]. For example, the sum of the annual budget of an intensive care unit and hospital overhead may be divided by the number of patient days to estimate an average cost per patient per day [34]. Bottom-up costing: An approach where cost components are valued by identifying resource use directly employed for a patient, resulting in patient-specific unit costs [14].
were not published in languages other than English, but any studies not reported in English were excluded from the review during screening. In addition, a hand search of key journals and the quoted references from the included articles was conducted to identify any additional studies. One author (B.D.) screened the titles and abstracts of all the citations identified from the search strategies, reviewed the full-text articles identified after screening and extracted the data from all included studies.

Data Items
Study design and population, types of bariatric surgery assessed, data collection methods and types of costs (i.e. cost components) included were extracted. Total costs of the procedure were extracted as well as the cost values associated with each individual component. Costs are reported here in the currency and price year originally listed in the included study. When the price year was not reported, it was assumed that the price year would be 1 year earlier than the year in which the study was published. To compare the cost of bariatric surgery between different studies, total procedural costs were also inflated to a 2016 price year. When prices were reported over multiple years, the most recent price year was used to inflate the total cost to make a conservative assumption. In terms of currency conversion for studies conducted in different countries, costs were adjusted for purchasing power parities (PPPs) (to adjust for price differences between countries, rather than just exchange rates, which do not take price differences into account) and converted to 2016 US dollar PPPs [16]. If total costs were reported in US dollars, but had been converted from the country's currency in which the study was conducted, the total cost estimate was first converted back to the original currency using the exchange rate provided in the article before inflating, adjusting and converting to 2016 US dollars.

Assessment of Accuracy, Reliability and Comprehensiveness
To provide an indication of the accuracy, reliability and comprehensiveness of the reported cost estimates, the inclusion of 'important' cost components based on criteria outlined by Ismail et al. [17] for conducting costing analyses of surgical interventions was assessed. Ismail et al. reviewed costing approaches for robotic surgeries, in general, and assessed 19 studies, three of which related to bariatric surgery [18][19][20], for their consideration of six criteria with the objective to create and validate a micro-costing methodology that could be used by surgeons and hospital administrators to evaluate the cost of implementing new surgical approaches. As standardised guidance regarding how to conduct a microcosting is limited [21,22], the methodology presented by Ismail et al. provides one of the only frameworks for the conduct of micro-costing studies and has specifically been designed for the evaluation of surgical interventions. The consideration of the six criteria presented by Ismail et al. was therefore thought to form a standard by which existing cost analyses could be measured. We have, however, also added an additional criterion (inclusion of overhead costs) to the six originally presented by Ismail et al. [17] as exclusion of such overheads could also affect the accuracy, reliability and comprehensiveness of cost estimates reported in the literature. A study was considered to include one of the 'important' cost components when a separate cost value could be identified for that individual component. The seven 'important' cost components included: 1. Cost, not charge data used in the analysis; 2. Operating room costs reported separate from hospital admission costs; 3. Medical device costs reported (e.g. endoscopy column, laparoscopic tower); 4. Personnel costs reported (e.g. surgeon, nurse, anaesthesiologist time); 5. Re-usable instrument costs reported (e.g. bowel graspers, surgical scissors); 6. Disposable instrument/consumables costs reported (e.g. needles, disposable staplers); and 7. Overhead costs reported.
Reporting of the methods used to identify cost components (gross and/or micro-costing) and value cost components (top-down and/or bottom-up costing) was appraised. When methods were not specifically reported, we assigned the relevant method based on the reported data collection description for resource items and costs. The inclusion of specific parameters based on standardised formulas [17] in the calculation of 'important' cost components (e.g. medical devices, personnel, re-usable and disposable instruments) was also evaluated. The parameters of interest included: 1. Capacity adjustment of personnel costs (e.g. adjusting salary for working days minus paid leave when determining a per-minute personnel cost); 2. Amortisation or depreciation of medical devices (e.g. allocating the acquisition cost of a device over its useful lifespan); 3. Maintenance fees for medical devices (in addition to the amortised acquisition cost); 4. Adjustments for medical devices shared across different procedures (i.e. laparoscopic tower may only be used 50% of the time for bariatric surgery);

5.
Sterilisation costs of reusable instruments (e.g. personnel time disinfecting and repackaging reusable instruments); and 6. Disposal costs of consumables (e.g. the waste management costs associated with consumables).

Synthesis of Results
Syntheses of the extracted data were performed to compare study characteristics and cost estimates across studies as well as combine individual study cost estimates into summary measures of total procedural costs. Important study characteristics and detailed cost estimates from the included studies were summarised in tabular format and used to evaluate the availability of cost estimates for different types of bariatric surgery, the methods used to collect the cost data and the specific cost components included in the analyses. Mean costs and standard deviations (SD) were calculated for all reported cost estimates together, estimates derived only from cost data (as charges do not necessarily reflect the actual cost of the resources consumed to deliver the surgery) and for different types of bariatric surgery. The inclusion of 'important' cost components was also synthesised in tabular format.

Study Selection
The search strategy identified a total of 998 citations. After removing duplicates (n = 499), 499 unique citations remained for title and abstract screening, which left 73 unique citations of interest for full-text review. After full-text review, 13 studies were selected for detailed review (Fig. 1).
A number of different types of bariatric surgery were costed, but no study assessed the cost of the SG procedure, perhaps as this is a relatively new procedure. One study reported costs for bariatric surgery, in general [32], two studies reported costs for different gastric banding procedures [25,31] and two studies reported costs for gastric bypass procedures [17,24]. The remaining studies compared the costs of two or more different procedures [18,23,[26][27][28][29][30]33].
Twenty-three total procedural cost estimates were reported in the 13 included studies, with a mean of $14,389 SD $6110 (range, $7423 to $33,541) ( Table 2). Excluding the five studies that used charge data to estimate total procedural costs left 14 estimates with a mean of $13,993 SD $5441. Five studies reported ten estimates of total procedural costs for different types of gastric bypass procedures with a mean of $13,496 SD $4171. Three studies reported four estimates of total procedural costs for different types of gastric banding procedures, with a mean of $15,237 SD $8556.

Inclusion of 'Important' Cost Components
The majority of studies used cost [17, 18, 24-26, 28, 29, 31] rather than charge data [23,27,32] in their analyses, with two studies [30,33] using both (Table 3). All the studies using cost data reported operating room costs separate from hospital admission costs, while only two [27,30] of the four studies using charge data reported these two costs separately. The remaining five 'important' cost components (medical device, personnel, re-usable instruments, disposable instruments and overhead costs) varied in their reporting. Nine studies reported costs for instruments [17, 18, 24-27, 29, 30, 33], but usually did not differentiate between re-usable and disposable [18,24,25,29,30]. Only four studies [17,18,25,29] included five or more of the seven important cost components in their analyses.

Methods and Parameters Used for Calculating 'Important' Cost Components
Only one study made specific reference to a method for identifying cost components [17]. Micro-costing methods were assumed to be most commonly employed to identify cost components, but methods of valuing components were not discernible in the majority of the studies. The inclusion of specific parameters in the calculation of 'important' cost components was also poorly reported. Only three studies amortised the cost [17,18] or accounted for depreciation [26] of medical devices over their life span and only two studies accounted for the cost of maintenance fees [17,18]. One of the studies reporting a combined operating room cost also noted that the laparoscopic equipment costs were amortised [29]. Only one study reported capacity adjusted personnel costs and accounted for sterilisation costs of reusable instruments [17]. No studies made adjustments for medical devices shared across different procedures or included the disposal costs associated with consumables.

Discussion
This paper presented a systematic review of cost analyses of a number of different approaches to bariatric surgery for the treatment of severe obesity. From the 13 studies included in the review, sources and methods of data collection were minimally reported, making it difficult to ascertain what methods were used to identify and value cost components. A number of different types of bariatric surgery were costed mainly from a US hospital perspective, including laparoscopic gastric bypass and gastric banding procedures, but no study reported a cost for the SG procedure. Some of the reported cost estimates were, however, for open procedures, procedures not commonly performed (e.g. vertical banded gastroplasty) or surgical techniques likely to be limited to certain providers (e.g. robot-assisted surgeries). This limits the generalisability and usefulness of the reported cost estimates for decision-making purposes as the majority of bariatric surgeries, worldwide, are performed laparoscopically using either the RYGB or SG procedure [2]. The inclusion of individual cost components in the total cost estimates varied widely, although the majority of the studies included hospitalisation and consumables/material/supply costs. Interestingly, only three studies explicitly considered surgeon/physician fees, despite the potential for these costs to drive differences in total procedural costs between the three procedures. Consideration of surgeon/physician fees in any future analyses will therefore be important, especially for comparing cost estimates between RYGB and SG where equipment/ instrument costs could be quite similar. The variation in included cost components can potentially be explained by the lack of a clearly defined care cycle or timeframe in which resource use and costs were measured for the majority of studies (only two studies defined a care cycle [23,30]), making it difficult to determine what cost components should be included.
Mean total procedural costs ranged from US$13,307 to US$15,237 depending on the types of studies included in the calculation. Excluding studies using charge data resulted in a lower mean total procedural cost (US$13,993 vs. US$14,389) and mean total procedural costs were observed to be lower for gastric bypass compared to gastric banding procedures (US$13,496 vs. US$15,237). These differences should, however, be interpreted with slight caution as these mean estimates are based on a small number of studies conducted in a number of different countries/settings and for a range of different procedures (e.g., open, laparoscopic and robotassisted). In comparison to the UK tariffs for bariatric surgery, the mean estimates reported in the literature are much larger [e.g. £4608 (US$6602) for 'Stomach bypass procedures for obesity-HRG code FZ84Z' and £4,028 (US$5,771) for 'Restrictive stomach procedures for obesity-HRG code FZ85Z']. The basis of the UK tariffs is, however, unknown and their use as a relevant indicator of the accuracy of the costs of bariatric surgery reported in our review may not be appropriate.
Overall, most studies in the review did not report accurate, reliable and comprehensive estimates of the total procedural costs as the inclusion of the 'important' costs components based on recommended costing methods for surgical procedures was poor, with no study including all of the components. Calculation methods were also poorly reported and usually did not account for recommended parameters when estimating costs. This is not surprising as detailed cost information for interventions is lacking in many clinical areas. LGB laparoscopic gastric banding, LGBP laparoscopic gastric bypass, LRYGBP laparoscopic Roux-en-Y gastric bypass, NA not applicable, In studies where the price year has not been reported, it was assumed that the price year would be 1 year earlier than the year in which the study was published Systematic reviews of costing studies of bariatric surgery are limited in the published literature. One relevant study was identified that reviewed cost approaches for robotic surgeries, in general [17], and despite the limited overlap in included studies, our review came to a similar conclusion, namely that costing studies related to surgical procedures (or specifically bariatric surgery) have not reported their methods transparently and largely do not consider 'important' cost components and parameters required for their estimation.

NCD
Furthermore, we have attempted to extract details of the reporting of specific parameters that would be required to generate appropriate cost estimates according to the formulas presented by Ismail et al. [17]. Just because a cost component has been included in an analysis does not mean that parameters required to estimate accurate, reliable and comprehensive costs have been considered. This point is highlighted in our review as even when important cost components were included, the methods and parameters used in their calculation were infrequently reported. Our review has also identified some additional criteria that should be reported to help improve the quality of cost estimates for bariatric surgery. Transparent reporting of the methods of identifying and valuing cost components should be provided, as the choice of method can have an impact on the magnitude of the cost estimates [13,14]. An explicit care cycle definition should also be provided, especially to differentiate between cost estimates that have and have not included costs that are incurred before and/or after the actual conduct of the procedure, such as the costs of nutritional and psychological evaluations, 6 to 12 months of medical weight management, re-admissions, postoperative complications, routine vitamin supplements and laboratory testing for the life of the patient after surgery. These additional costs may be significant, but from the available literature, it is not clear if such costs were considered due to the lack of care cycle definitions.
Our study does, however, have limitations. First, detailed costings of bariatric surgery may have been conducted by certain healthcare providers, but not available in the literature. This is possibly due to the inclusion of sensitive pricing information (e.g. discounts negotiated with manufacturers of certain equipment) and, therefore, unlikely to be publically available.
Furthermore, our review only identified a small number of studies, with the majority conducted in a US hospital context. This makes it difficult to generalise the results of our study to other settings/jurisdictions, as different equipment, materials and personnel may be involved across different sites and countries. Caution should, therefore, be made when interpreting the mean cost estimates presented in our review in a local context. Overall, our review indicates that there is a need for up-todate costings of the three most common bariatric procedures (RYGB, AGB and SG). To ensure these costs are collected in a consistent manner, micro-costing methods have been identified as the most appropriate approach. We plan to conduct such a micro-costing study across a number of hospitals offering bariatric surgery within the NHS in England as part of the BBS study [5,10]. Our study has been designed to consider all the 'important' cost components outlined in this review, will report the parameters involved in their calculation in a transparent manner and explicitly define the cycle of care to ensure inclusion of all relevant cost components. We will, therefore, be able to determine accurate, reliable and comprehensive estimates of the cost of bariatric surgery.