What Are the Uses and Limitations of Time-driven Activity-based Costing in Total Joint Replacement?

Abstract

Background

With increasing emphasis on value-based payment models for primary total joint arthroplasty (TJA), there is greater need for orthopaedic surgeons and hospitals to better understand the actual costs and resource use of TJA. Time-driven activity-based costing (TDABC) is a methodology for accurate cost estimation, but its application in the TJA care pathway across institutions/regions has not yet been analyzed.

Questions/purposes

In this systematic review of studies applying TDABC to primary TJA, we investigated the following: (1) Is there variation in TDABC methodology and cost estimates across institutions? (2) Is a standard set of direct and indirect costs included across studies? (3) Is there a difference in cost estimates derived from TDABC and traditional hospital cost-accounting approaches? and (4) How are institutions using TDABC (process and outputs) with respect to the TJA care pathway?

Methods

A comprehensive search strategy was developed that included the keywords “TDABC,” “time-driven activity-based cost,” “THA,” “TKA,” “THR,” “TKR,” and “TJR” in the PubMed/MEDLINE, EMBASE, Web of Science, Ovid SP, Scopus, and ScienceDirect databases for articles published between 2004 and 2018 as well as extensive hand searching and citation mining. Relevant studies (n = 15) were screened to include THA or TKA as the focus of the TDABC model, full-text articles, TDABC-based cost estimates for TJA, and studies written in English (n = 8). Due to the heterogeneity of outcomes/methodology in TDABC studies involving TJA, quality assessment was based on each study’s adherence to the seven steps delineated by Kaplan et al. in their original publication introducing TDABC in health care.

Results

There was substantial variation in TDABC methodology (especially in scope), adherence to the seven steps of TDABC, and data collection. Only five of eight studies incorporated indirect costs into their TDABC calculation, with notable differences in which direct and indirect expenses were included. TDABC-based cost estimates for TJA ranged from USD 7081 to USD 29,557, with variation driven by the TJA timeframe and whether implant costs were included in the costing calculation. TDABC was most frequently used to compare against traditional hospital accounting methods (n = 4), to increase operational efficiency (n = 4), to reduce wasted resources (n = 3), and to mitigate risk (n = 3).

Conclusions

TDABC-based cost estimates are more granular and useful in practice than those calculated via traditional hospital accounting; however, there is a lack of standardized principles to guide TDABC implementation (especially for indirect costs) due to institutional and regional differences in TDABC application. Although TDABC methodology will likely continue to vary somewhat between studies, standardized principles are needed to guide the definition, estimation, and reporting of costs to enable detailed examination of study methodology and inputs by readers.

Level of Evidence

Level III, economic and decision analysis.

Introduction

The expansion of value-based reimbursement and alternative payment models has made it imperative that physicians and hospitals better understand resource use and the costs of care. Currently, traditional hospital cost accounting methods, such as the ratio of cost-to-charges and relative-value units, have customarily been used to estimate the costs of care [16]. The ratio of costs-to-charges uses a set ratio for costs to charges and applies that ratio to the total charges for all billed procedures or services [3]. The relative-value-units approach assigns to each procedure or service a number of relative value units that are based on the estimated skill required and personnel or equipment involved. Total costs are divided by the total relative value units to arrive at dollars of cost per relative value unit. Although these approaches simplify cost calculation, they have their limitations. The ratio of costs-to-charges method does not consider the differences in time and personnel effort, while relative-value-units fails to account for a complete set of direct and indirect costs of care. Neither approach accounts for time-based differences among individual instances of the same procedure [14].

In 1988, Cooper et al. [6] proposed activity-based costing, in which indirect and overhead costs (such as nonclinical wages) are divided into cost pools that are then assigned to specific activities (such as revenue cycle billing) based on staff interviews. Each activity has a specific cost driver rate, a quantifiable element that drives the cost of the activity (for example, the number of billable claims or codes for billing overhead activity). Each service in an outpatient practice generates an average number of claims or codes that must be billed, leading to an expected nonclinical wage cost from revenue cycle billing that is allocated to each procedure or service performed. While activity-based costing has the advantage of accounting for various activities that drive indirect or overhead costs and more accurately assigning them to specific procedures or services, the model is difficult to use on a large scale because it requires updating whenever there is a change in costs for a given service. This requires reinterviewing of frontline and expert staff to reevaluate and update the cost pools and cost driver rates of activities within the care pathway over time. Activity-based costing also introduces potential bias by relying on the subjective opinions of individuals with the natural inclination to demonstrate 100% productivity all the time. In reality, productivity levels are probably closer to 80% to 85%, accounting for factors such as presenteeism (nonproductive time at work). Activity-based costing models also require substantial resources for data collection and analysis, and can create a large burden on the systems that use them [10].

To address the limitations of established cost-accounting methods, Kaplan et al. [10] proposed time-driven activity-based costing (TDABC) in 2004. A more customized approach to TDABC in health care proposes a seven-step method that can be used to estimate the cost of care for an individual patient treated along a care pathway for a particular condition [11]. In this method, clinical and nonclinical personnel provide input to generate process maps of all activities needed for a specific service or procedure. The TDABC team, as opposed to frontline employees, estimates the practical capacity (that is, the actual productive time) of each capacity-supplying resource (employee and equipment) and the average time required for each resource activity, based on surveys or direct observation. This “bottom-up” approach assigns the time of capacity supplying resources to activities that are then allocated to services or procedures. The model allows new activities to be incorporated into the process without overhauling the whole model.

Numerous studies [2, 7-9, 14] have evaluated TDABC for patients undergoing total joint arthroplasty (TJA) covering the index admission through 90 days after surgery. These studies reported large differences in estimates of cost and resource use for TJA, but it is unclear how much of this variation is because of circumstantial differences unique to a region or institution or because of differences in methodology. To better understand the drivers behind the heterogeneity of TDABC results, we divided the major differences in TDABC implementation into four different categories based on our preliminary analyses of these studies: (1) methodology, (2) indirect/direct cost inputs (3) cost estimates (4) application of outputs.

This systematic review therefore aimed to answer four questions: (1) Is there variation in TDABC methodology and cost estimates across institutions? (2) Is a standard set of direct and indirect costs included across studies? (3) Is there a difference in cost estimates between TDABC and traditional hospital cost-accounting methods? and (4) How are institutions using TDABC (process and outputs) with respect to the TJA care pathway?

Patients and Methods

Protocol, Eligibility Criteria, Information Sources, and Search Strategy

A review protocol was developed for this study but not registered. We included studies with THA or TKA as the focus of the TDABC model, full-text articles (excluding abstracts, conference proceedings, etc.), cost estimates for TJA using the TDABC model, and studies written in English with no limitation on geography. The search strategy specifically included the keywords “TDABC” or “time-driven activity-based cost” and “THA,” “TKA,” “THR,” “TKR,” or “TJR.” We searched for these keywords in the PubMed/MEDLINE, EMBASE, Web of Science, Ovid SP, Scopus, and ScienceDirect databases for articles published between 2004 and 2018. We also conducted hand searching and citation mining to find TDABC studies focused on TJA; the last search was conducted in December 2018. Relevant articles (n = 101) were independently screened by two reviewers (SP, TK), based on relevance to TDABC and orthopaedic procedures (Fig. 1). A third reviewer was available for consultation in case of disagreement, although this did not occur.

Fig. 1.

This figure depicts the method we used to select studies for the different phases of article review, according to the Preferred Reporting Items for Systematic Review and Meta Analysis strategy. We developed a comprehensive search strategy that included the keywords “TDABC” or “time-driven activity-based cost” and “THA,” “TKA,” “THR,” “TKR,” or “TJR” in PubMed/MEDLINE, EMBASE, Web of Science, Ovid SP, Scopus, and ScienceDirect databases. Abstracts were screened for relevance to TDABC and orthopaedic procedures. Full-text articles were excluded if they did not meet the following criteria: total THA or TKA as the focus of the TDABC model, availability of full-text articles, inclusion of cost estimates for the TJA procedure using TDABC, and studies written in English.

Study Selection

Of 101 screened studies, only 15 applied TDABC to orthopaedic procedures and were further analyzed. Studies needed to include, at a minimum, some mention of how the TJA care pathway was defined, sources for cost and time inputs, and any assumptions driving time-driven costs. Of these 15 studies, seven were removed because they did not specifically focus on patients undergoing primary THA or TKA, leaving eight that ultimately met the inclusion criteria.

Data Collection Process and Variables

Two reviewers (SP, TK) independently extracted data on institution and sample size, sample size for time-based inputs, participation in bundled payment programs, scope of clinical activities included in the TDABC model, and data sources for inputs (Table 1). Studies that met the inclusion criteria were then divided into three comparison groups based on the scope of clinical activities, which included the index admission, index admission plus 30-day period, or index admission plus 90-day period. The studies were screened against the seven steps for TDABC outlined by Kaplan et al. [11] (Table 2). Data on direct and indirect cost inputs for each study were collected, categorized, and compared among subgroups. If it was not explicitly specified, we assumed that the input was not included. Direct cost inputs were categorized into hospital and postacute care costs. Direct cost categories were separated based on the criteria set by Navathe et al. [13], and indirect cost categories were determined by collecting information on indirect cost inputs from each study. We compared total cost estimates for THA and TKA among the included studies. Direct costs were defined as clinician labor, equipment, implants, consumables, and lab testing that could be directly assigned to a service or procedure. Indirect costs were defined as fixed or periodic costs of support resources necessary to supply a service or procedure [11]. This includes expenses for nonclinical salaries (such as billing and scheduling), space, sterilization and storage of equipment, information technology, and other activities or departments that support and enable patient care (Table 3).

Table 1.

Characteristics of the studies included in this systematic review

Table 2.

Assessment of TDABC in the included studies relative to the seven-step Kaplan method [11]*

Table 3.

Direct and indirect cost inputs for TDABC estimates*

Study Appraisal

In the seminal article outlining TDABC and applications in health care, Kaplan et al. [11] provide a seven-step framework for calculating time-driven costs. We used the latter as a proxy for measuring TDABC study quality (from a methodological perspective), and we assumed a given step was not performed unless explicitly stated. Although all studies attempted to follow Kaplan et al.’s seven-step TDABC approach, only five studies met all seven steps (Table 2).

Data Analysis

We made qualitative comparisons across the studies to assess for similarities and differences in data collection methodology, direct and indirect costs inputs, TDABC outputs, and application of findings. Descriptive and statistical analyses were used to compare final cost estimates for THA and TKA between institutions, recognizing these differences are partially driven by differences in TDABC costing methodology. Using data across included studies, we computed a mean TDABC-based cost for TJA according to the three time frames: index admission only, index admission plus 30 days postsurgery, and index admission plus 90 days postsurgery (Table 4).

Table 4.

Comparison of TDABC estimates by study and procedure time period

Study Characteristics

Among the eight included studies, six were from US-based institutions [1, 7-9, 13, 14], one was from the UK [5], and one was from Denmark [2], with a wide range of institutional sample sizes (range, 1-29 facilities). Seven included studies [1, 5, 7-9, 13, 14] analyzed academic hospitals and three analyzed community hospitals [2, 9, 13] (two analyzed both [2, 9]). Three studies [7, 8, 13] involved hospitals that were already affiliated with a bundled payment model for primary TJA (Table 1). Six studies [1, 2, 7, 8, 13, 14] reported TDABC estimates for both THA and TKA, one [5] reported TDABC estimates for only TKA, and one [9] did not report a TDABC-based cost estimate. Instead, this study compared TDABC costs for TKA across 29 facilities, with the median cost estimate valued at 1.00 [9] (Table 1).

Results

Variation in TDABC Methodology Across Institutions

Each study used different data gathering methods to calculate time-based inputs of capacity supplying resources. Six of eight studies used direct observation [1, 2, 5, 8, 13, 14], two interviewed hospital staff [1, 14], two used electronic hospital records [2, 14], and two did not specify the method that was used [7, 9] (Table 1). Additionally, one study did not report using process maps for constructing the TDABC model [13]. The studies were further subdivided according to which section of the full TJA pathway (index admission plus 90 days) was analyzed. Five studies analyzed the index admission plus 90 days postsurgery [2, 7, 8, 9, 14], one analyzed the index admission plus 30 days postsurgery [13], and two analyzed the index admission only [1, 5] (Table 3).

Among studies applying TDABC to the index admission plus 90 days postsurgery [2, 7, 8, 14], the estimated TDABC costs for THA and TKA ranged from USD 14,924 to USD 29,557; among those with less than the 90-day timepoint [1, 5, 13], the total TDABC estimates ranged from USD 13,322 to USD 21,208, with one study [5] reporting only the cost of TKA (USD 7081) (Table 4).

Direct and Indirect Cost Inputs Included in the TDABC Models

All studies looking at index admission and 30- to 90-day followup [2, 7-9, 13-14] included direct costs associated with the surgical stay (that is, consumables, operating room, implant, and room and board), including some postacute care costs (skilled nursing facility, inpatient rehabilitation, home health services, and physical therapy) occurring after hospital discharge (Table 1). The largest and most common direct cost categories included implant [1, 5, 7-9, 13, 14], surgical supplies [1, 2, 5, 8, 9, 13, 14], medications [1, 2, 5, 8, 9, 13, 14], and laboratory testing [2, 5, 7-9, 13, 14] (Table 3). Of the six studies that reported postacute care costs [2, 7, 9, 13, 14], five incorporated outpatient visit expenses into their costs [2, 8, 9, 13, 14]. Seven studies did not include the cost of durable medical equipment costs [1, 2, 5, 7-9, 14], and six did not include costs of emergency department visits related to the index surgery [1, 2, 5, 8, 9, 14].

Five studies reported incorporating indirect cost into their TDABC calculation [1, 5, 7, 8, 14]; however, only four [1, 5, 8, 14] listed the exact items included under “indirect cost.” Among the four studies that specified indirect costs, the most prevalent inputs were billing [1, 5, 8, 14], information technology [1, 5, 14], maintenance and housekeeping [1, 5, 14], and hospital administration [1, 14]. Most studies did not include human resources [1, 14], nonclinical salaries [5, 8], and indemnity [5] in their indirect cost calculations.

TDABC Versus Traditional Hospital Cost-accounting Estimates

Two studies [1, 14] additionally reported traditional hospital cost accounting-based 90-day period costs of USD 24,356 and USD 25,782 for THA and TKA, respectively, compared with TDABC-based estimates of USD 13,322 and USD 14,123 for THA and TKA, respectively. In these studies, TDABC-based estimates were USD 10,858 lower for THA and USD 11,034 lower for TKA than those for traditional hospital cost accounting.

Uses of Insights and Outputs from TDABC Models

The most common uses for TDABC were to increase operational efficiency [1, 2, 5, 13], compare against traditional hospital accounting methods [1, 2, 5, 7], reduce wasted resources [1, 2, 5, 7], and mitigate risk [7, 8, 14] (Table 5). Studies that analyzed a shorter period than index admission plus 90 days postsurgery were more likely to use TDABC for reducing resource waste and increasing operational efficiency [1, 5, 13], while those that analyzed the full 90-day period more often used TDABC estimates for comparison against reported costs, mitigating risk under bundled payment models, and informing reimbursement negotiations [2, 7, 8, 14].

Table 5.

Use of TDABC processes and outputs by study

Discussion

TDABC is increasingly implemented as a more efficient, accurate method for understanding the true cost of care delivery. Beyond less-intensive use of postacute care facilities—the leading strategy for cost reduction used by surgeons and hospitals participating in bundled payments—TDABC may help identify waste and redundancies to improve efficiency within the TJA care pathway via process mapping. [4, 12]. To date, beyond a high-level seven-step framework and a handful of case examples, there has been minimal guidance to authors/researchers around the best practices for TDABC methodology, indirect/direct cost inputs, cost estimation, and applications of outputs from TDABC implementation. This systematic review therefore aimed to answer the following questions: (1) Is there variation in TDABC methodology and cost estimates across institutions? (2) Is a standard set of direct and indirect costs included across studies? (3) Is there a difference in cost estimates between TDABC and traditional hospital cost accounting methods? and (4) How are institutions using TDABC (process and outputs) with respect to the TJA care pathway?

Limitations

The large range in TDABC-based TJA cost estimates across studies is partially driven by methodological choices by the author/institution (that is, scope of the care pathway analyzed, indirect/direct cost inputs included, etc.); at the same time, differences between institutions regarding efficiency (such as, the expected time for activities/processes, staff required) could ultimately contribute to differences as well. Since our analysis was restricted to the inputs, process maps, and TDABC calculations provided in the manuscript of included studies, it is challenging to comment on overall efficiency differences between institutions given the data available. However, after standardizing for TDABC methodology differences, a comparison of cost estimates may highlight specific efficiency differences between institutions at different points/activities in the TJA care pathway. Additionally, TDABC models are subject to institutional and regional variations in care pathways, clinical practices, and pricing (especially for direct consumable and clinician salaries), resulting in differences in cost inputs and estimates. These limitations underscore the importance of greater transparency and standardization in the reporting of TDABC modeling methodology in TJA, including handling of indirect cost estimates.

The results of this review suggest that TDABC-based cost estimates, although variable across studies, are consistently lower, more granular, and more useful in practice than those calculated with traditional cost accounting methods. Insights from TDABC modeling were most commonly applied to increase operational efficiencies, reduce waste, enable accurate cost comparisons, and mitigate risk under bundled payments. Although the theory and principles of TDABC are well defined, the current study demonstrates how authors and institutions have interpreted TDABC in many ways, especially with respect to following the Kaplan seven-step method [11], scope of direct and indirect costs, and the portion of the TJA care pathway to which TDABC was applied. While some amount of variation in TDABC methodology and outputs is expected, it must be balanced by following the core elements of TDABC and providing a granular view of the underlying methodology (including how inputs were collected and defined) so readers and providers can interpret findings accordingly. Additionally, publication bias may inflate the utility of TDABC as only those articles that found and demonstrated utility of applying this methodology are likely to get published.

Variation in TDABC Methodology Across Institutions

Kaplan et al. [11] provided a seven-step framework for calculating time-driven costs of treatment (Table 2), which was referenced in all analyzed studies. However, only five of eight studies followed all steps, with considerable variation in how they were performed [1, 5, 7, 8, 14]. This demonstrates that while the underlying principles of TDABC are well understood, actual implementation is subject to interpretation based on authors and providers, focus of the analysis, and institution-specific constraints [1, 5, 7, 8, 14]. Among the seven studies [1, 2, 5, 8, 9, 13, 14] that reported personnel (clinical salaries) and equipment costs, two used national benchmarks in addition to internal accounting data [1, 8], while all the others used only internal accounting data from varying sources. Furthermore, there was heterogeneity in generating estimates for time spent on clinical and nonclinical activities in the TJA pathway that ranged from direct observation [1, 2, 5, 8, 13, 14] to collecting electronic health record data [2, 14] to interviewing staff [1, 14] to video recording of activities [8]. These factors may have contributed to the large differences in TDABC cost estimates among the included studies.

The scope of the care pathway included in the TDABC calculation also varied substantially. As expected, studies that analyzed the cost of the initial admission to 90 days postsurgery reported a higher cost than did those that analyzed a more limited section of the pathway. Inclusion of postacute care costs in the 90-day period is critical to assessing the downstream effects of decisions made during the index admission, including discharge to a skilled nursing facility or inpatient rehabilitation facility versus home care, choice of a nonhome facility, and prophylaxis against potentially preventable complications (such as deep vein thrombosis, surgical site infection, and implant-related infection). Although this introduces greater complexity, it provides a more complete financial picture of care.

Direct and Indirect Cost Inputs Included in the TDABC Models

Three studies did not report including any indirect costs in their TDABC calculation [2, 9, 13]. Among the five studies that did [1, 5, 7, 8, 14], one applied a blanket assumption that indirect costs consist of 60% of direct costs [1], another estimated indirect costs to be 16.5% of the total cost of the TJA episode [14], and a third assumed overhead to be 43.8% of the net difference in revenue minus the cost of providing services [5]. None of these studies provided compelling evidence to support the use of any particular set of assumptions. However, one study justified excluding indirect costs, stating that they accounted for less than 10% of personnel and supply costs [9]. Although TDABC costs will likely never be completely standardized, standardized principles are needed to guide the definition and estimation of indirect costs. Given that billing, information technology, hospital administration, and maintenance costs were most commonly recorded and likely comprise the largest sources of indirect cost, we propose a standardized indirect cost set that includes these four categories at a minimum (Table 3).

TDABC Estimates Compared with Traditional Cost-accounting Methods

TDABC estimates for TJA varied considerably across studies (USD 2550–USD 33,216) [2, 13]. This was expected—given the variation in TDABC methodology, data sources, and cost categories that were included—and highlights the importance of examining study methodology granularly before taking TDABC cost estimates at face value. For example, the study with the lowest cost was conducted in Denmark and did not include implant costs in the final TDABC estimate [2]. Among the remaining six studies that included the final TDABC cost estimates [1, 5, 7, 8, 13, 14], five were US-based and one was UK-based [5]. Variation in the cost of orthopaedic procedures across the US and in other countries is well documented [15]. In the study reporting the highest cost for THA and TKA, Navathe et al. [13] additionally recognized that including patients with major comorbidities (such as myocardial infarction, pneumonia, aortic stenosis, and renal or liver transplants) affects cost estimates because of increased length of stay, discharge to a skilled nursing facility or inpatient rehabilitation facility, and need for more intensive followup once patients go home [13]. The average cost of noncomplex primary THA or TKA was USD 21,208, while complex primary THA and TKA had an average cost of USD 33,216.

Only two studies in our analysis compared traditional cost accounting and TDABC estimates [1, 14]. In both of those studies, traditional cost accounting resulted in higher cost estimates than did TDABC. This difference is likely driven by a combination of factors, including that traditional cost accounting attributes 100% of employee salaries to procedures, ignoring the reality that 15% to 20% of employee time is consumed by nonproductive activities. Thus, while comparing TDABC and traditional hospital cost accounting can be instructive for an institution (such as when negotiating for reimbursement or attempting to understand which services and procedures are profitable), publishing these numbers out of context may be misleading.

Uses of Insights and Outputs from TDABC Models

In addition to providing more accurate costs for patient encounters, the authors and institutions in our analysis used TDABC to increase operational efficiency [1, 2, 5, 13], to reduce resource waste [5, 9, 13], and to mitigate risk when using alternative payment models [7, 8, 14]. By deconstructing complex services through process mapping, organizations may better understand the major cost drivers within a service. By focusing on the processes that contribute to a patient's experience [8], unnecessary or low-value processes can be eliminated and activities can be staffed by appropriate personnel [1, 8, 14]. Accurate estimates based on the activities and staff at a particular site can also facilitate meaningful comparisons between clinical teams or sites to further drive best practices [2]. TDABC is less burdensome from a resource-use perspective when compared with activity-based costing, since changes in processes, time, or staff requirements can be easily updated in TDABC models.

Lastly, TDABC provides a valuable tool to minimize risk when entering value-based payment negotiations by informing where potential risks lie via process mapping and by providing a check against traditional cost accounting figures. As an example of the latter, one study noted how comparison to TDABC highlighted that traditional cost accounting tends to overestimate certain cost categories (nonphysician personnel, space and equipment) while TDABC (if applied too narrowly) can ignore relevant infrastructure expenses not directly dedicated to the care of a single patient but that still should be included as institutional overhead for the purposes of bundled payments [14]. By tracking time inputs on a per patient level, hospitals are able to identify patient characteristics associated with higher TDABC costs. This can be used in contract negotiations with payers to inform risk-adjusted payments or create customized cost and payment estimates through analyzing the payer population.

Conclusions

There are several challenges to successfully applying TDABC to primary TJA, the most prominent of which is a lack of standardized principles to guide TDABC implementation (especially for indirect costs) and reporting to enable granular examination of study methodology and input. Despite this, TDABC-based cost estimates are more detailed and useful in practice than those calculated with traditional cost accounting methods. The insights from TDABC modeling have been used by various authors and institutions to improve operational efficiency, reduce waste, enable accurate cost comparisons, and mitigate risk under bundled payments for TJA. These insights are meant to inform institutions looking to (1) perform TDABC in a complete, rigorous manner, (2) report TDABC methodology and outputs in a format that supports comparability with peer institutions and reproducibility via a standardized framework, and (3) understand how TDABC outputs can be adapted towards an institution’s purposes, especially for those looking to transition towards value-based care.