17 min read

Cost measurement in value-based healthcare: a systematic review

Cost measurement in value-based healthcare: a systematic review

Cost measurement in value-based healthcare: a systematic review

STRENGTHS AND LIMITATIONS OF THIS STUDY

  • Using two independent reviewers, this systematic review analyses all value-based healthcare studies that have to date measured costs to provide a comprehensive comparison of cost measurement methods.
  • This research operationalises the benefits of cost measurement to practitioners by identifying four mechanisms through which cost information facilitates value-based healthcare.
  • By comparing the methods used in literature to collect cost information, this study identifies three best practices for practitioners and researchers.
  • By limiting the search to studies labelled as ‘value based’ in their title or abstract, this review overlooks studies that measure healthcare costs but do not explicitly relate this to value-based healthcare.
  • The included studies may have achieved value-based healthcare without reporting this explicitly, and therefore, may be overlooked when answering Research Question 2.

Introduction

To make sound value-based decisions in healthcare, hospital practitioners and healthcare providers require patient-level information on the costs incurred and outcomes achieved in hospitals and other healthcare organisations.1 This will enable care providers to steer towards better patient-reported outcome measures, better patient-reported experience measures and clinical outcomes at equal or lower cost.2 With detailed cost and outcome information, care paths can be continuously optimised.3 Consequently, value-based healthcare (VBHC) is considered one solution to the financial pressures our global healthcare system places on managers and administrators1 4 5 based on its promise to streamline care by focusing on desirable outcomes. In addition, hospitals can benefit from cost information by gaining insight into the sources of costs that can then guide cost-containment strategies. Therefore, cost information may facilitate process and quality improvement initiatives pursued by management.6–10 Furthermore, insight into patient-level or treatment-level costs enables hospitals to negotiate appropriate prices with insurance firms, especially given the trend towards new payment models and away from fee-for-service payments.11 12 Finally, it is suggested that such treatment-level cost information enables market-based competition among hospitals based on outcomes and prices.13

Considerable research has addressed the outcome side of Porter’s value equation.14 This value equations suggests that healthcare should pursue ‘value’, where value is defined as desirable and relevant patient level outcomes divided by the costs of delivering care.1 2 Many studies have measured patient-level outcomes from both the patient perspective (eg, patient-reported outcome measures, patient-reported experience measures) and clinical outcome perspective.15 16 Less is known about the cost side of this equation. Often, the term ‘cost’ is conflated with the price paid by insurance firms or patients to the hospital.17 18 However, prices do not reflect the costs incurred by hospitals.6 19–21 Prices paid by insurance firms are negotiated sums that include profit margins for both the insurer and the hospital.22 They are also impacted by political factors, such as the hospital-payor mix23 that refers to the range of private and public insurance schemes that make up the hospital’s income stream. Finally, fee-for-service payments fail to account for patient-level differences in required care. Therefore, reimbursements are considered a poor indicator of costs.

Some authors argue for time-driven activity-based costing (TDABC) as the ‘gold standard’ of cost measurement in healthcare organisations.3 5 15 TDABC, in a fine-grained way, matches direct and indirect costs to activities based on the time an activity takes. A care path is made up of many activities, each generating costs. The costs of a care path can thus be calculated by first identifying all costs relevant to each activity, and then summing these costs across the activities.22

Although the research is growing and results are promising, there is relatively little empirical evidence to support TDABC being the best costing method to enable VBHC since studies rarely compare methods, and often simply use whichever system the investigated hospital or care provider uses. Costing methods differ by how they allocate indirect costs to products or services.24 Moreover, indirect costs cannot causally be attributed to patients, and therefore, need to be appropriately allocated. An example of such indirect costs are the salaries of administrative personnel such as the front office staff who welcome patients, coordinate schedules and manage equipment. While some costing methods ignore this (eg, direct costing), other methods average indirect costs across days or months, or systematically allocate them to patients. These methods range from imprecise to fine-grained, with TDABC towards the fine-grained end of the scale. This insight is particularly relevant to healthcare since indirect costs are high. The most fine-grained method is known as activity-based costing (ABC) and allocates indirect costs based on actual units of resources used per activity. In comparison, TDABC allocates indirect costs based on a per-minute cost, making it considerably easier to implement. Costing methods that ignore the indirect costs of a care path underestimate the true costs of the care delivered.

Previous systematic reviews have found that TDABC was able to facilitate VBHC, often highlighting cost savings as a result but without comparing it to alternative methods.3 4 15 Therefore, we do not know how TDABC compares to other cost measurement methods currently in use. While TDABC may be able to facilitate VBHC,5 15 it is unclear how its benefits compare to other costing methods. For these reasons, the cost side of the value equation remains unclear. To address this challenge, we pose two research questions:

RQ1: Which costing methods are currently being used by practitioners to facilitate VBHC?

RQ2: What are the consequences of applying a specific costing method in VBHC? These consequences may include whether the method enables a cost reduction with equal or better health outcomes or provides sufficient information to further improve a care path.

This comprehensive review draws on management accounting literature to categorise costing methods reported in empirical VBHC literature published over the last two decades (1 January 2003 to 1 January, 2022) into cost measurement methods defined in the literature,24 such as direct costing and absorption costing. Compiling studies in this way revealed four ways through which cost information facilitates VBHC and three best practices.

Materials and methods

Literature search strategy

To identify relevant studies, we systematically searched four major databases: Embase, Medline, Web of Science and CINAHL EBSCOhost. Our search string (online supplemental appendix) was developed by assessing previously identified relevant papers and was designed to catch all studies that address VBHC and measure the costs of an intervention, care path or treatment by including the following specific terms:

*cost*, microcost*, macrocost* AND [meaning in combination with] value-based, value based, OR valuebased

Initial search string testing showed that restricting the search to the phrase “value-based healthcare” excluded too many relevant studies because authors use phases such as “value-based perspective” or “value-based equation” when referring to VBHC. Conversely, the term “value” was too broad and yielded more than 40 000 mostly non-specific results. By using wildcard terms indicated by stars we included many variations on the term ‘cost’.

Eligibility criteria, record selection and data collection

We limited ourselves to peer-reviewed empirical research that measured or estimated costs in a VBHC context. All the inclusion criteria and variables extracted are detailed in online supplemental appendix. The following variables, inspired by Porter2 and the cost measurement methods defined in the accounting literature, were noted:

  • Cost types included (direct vs indirect).
  • Cost perspective (provider, payer, patient).
  • Portion of the care path costed (full, partial).
  • Cost measurement method used (as labelled by authors, verbatim).
  • Cost measurement categories based on accounting definitions, for example, direct costing, absorption costing, step-down allocation and other recognised methods.24
  • Consequences of the costing information generated.

Patient and public involvement

This study did not involve patients or the public in designing, executing or reporting the research.

Results

Record selection

Our four-person (ML, PP, HvE and KA) research group identified 3275 relevant papers, of which 1930 remained after removal of duplicates. We conducted a trial screening of 30 papers to test and further specify screening criteria. The screening process comprised two rounds, as shown in figure 1. In round 1, ML and PP screened the titles and abstracts independently. When there was uncertainty about the eligibility of a paper, it was retained for full-text screening following Bramer.25 We accepted 674 studies based on titles and abstracts, with a Cohen’s kappa inter-rater reliability score of 0.78, indicating substantial agreement.26Figure 1

PRISMA flow chart depicting the screening, exclusion and inclusion processes with two reviewers. PRISMA, Preferred Reporting Items for Systematic Reviews and Meta-Analyses; VBHC, value-based healthcare.

In round 2, both ML and PP screened the full text of all 674 studies independently. Of these, 215 studies were seen as relevant for RQ1, with a Cohen’s kappa of 0.76 between ML and PP. HvE was included in any resolution discussions needed. Finally, we assessed whether each paper discussed if or how the costing information facilitated VBHC (RQ2), yielding 49 instances where the costing method facilitated VBHC. This review was not registered.

Descriptive characteristics

An overview of the included studies is provided in table 1. Our earliest study is from 2005, with an upsurge in studies from 2017 onwards. Just under half (n=98, 45.6%) of studies were published in the last 2 years. An overwhelming majority are from the USA (n=178, 82.8%). Europe is the second most common continent with 22 (10.6%) studies of which 9 (4.2%) relate to Dutch healthcare.Table 1

Characteristics of value-based healthcare studies that measure costs (n=215)

The three largest medical specialty groups represented are surgical (n=99; 46.0%), oncology (n=37; 17.2%) and paediatrics (n=19; 8.8%). A complete list of the 215 studies included is summarised in online supplemental appendix. Extracted data are available in online supplemental file.

Which cost measurement methods are currently being used to facilitate VBHC?

To answer RQ1, we look at how costs were measured. A summary of our findings is presented in table 2. The literature contains many overlapping and contradictory terms, as ‘costs’ can refer to insurer costs, reimbursements, hospital costs, or patient costs. About half of the studies (n=110, 51.2%) take a provider perspective, with costs calculated for the hospital or care facility. Many studies use charges or payments because hospital cost data are unavailable, considering charges to be a relevant proxy. Some studies use terms such as ‘costs’, ‘charges’, ‘prices’, ‘payments’ and ‘reimbursements’ interchangeably, making it difficult to differentiate.17 18 27–30 For example, Jain et al17 stated, ‘The terms reimbursement, cost, and payment have been used interchangeably throughout the text to represent actual amounts paid by insurers.’ Similarly, Robles et al30 explained, ‘Total hospital charges were used in this standardised costing analysis. Hospital charge data provides a relative measure of the ‘cost’ of episodes of care, as actual cost data are generally not ascertainable in the healthcare setting.’ When calculating costs using TDABC, Ahluwalia et al31 called these costs ‘prices.’ To try to address this confusion, some recent studies refer to provider costs as the ‘true cost’ of care.6 7 9 19 Some studies that compare several cost types19 20 also differentiate ‘traditional hospital accounting’ costs from ‘true costs’ calculated with TDABC.6 9 20 32 33Table 2

Characteristics of costing methods in value-based healthcare

We categorised studies based on the cost types included. Both direct and indirect costs were considered in 177 (81.9%) studies, while 28 (13.0%) papers only included direct costs.

Next, we looked at whether costs were calculated for a complete care path. We found 64 (29.8%) studies that measured costs for a full care path, of which 16 (7.4%) refer to full surgical episodes and label them as such without considering all the presurgical or postsurgical costs. The remaining 86 (42.1%) measure costs of a partial care path.

Table 3 categorises studies based on the costing method used. In those papers measuring costs within a care provider, we identified two clear categories that were in line with the management accounting literature.24 The first is ‘direct costing’ (n=23), where direct costs of care are summed and indirect costs ignored. This implies that, if costs cannot be causally attributed to the treatment of a specific patient, they are not considered and hence overlooked when making managerial decisions.24Table 3

Overview of cost measurement methods used in value-based healthcare

The second category of studies considers both direct and indirect costs and uses ‘absorption costing’, whereby indirect costs are allocated to patients based on an allocation key (a type of formula used for allocating indirect costs).31 These studies include but are not limited to TDABC (n=31) and ABC (n=7), where costs are allocated to individual care activities (such as a consultation or treatment step). The remaining absorption costing papers (n=47) also consider direct and indirect costs but do not report how indirect costs are allocated to activities. In the absorption costing studies, authors may state that cost information was calculated based on diagnosis-related group costs, microcosting, bottom-up clinical costing or hospital accounting systems not further classified. A full list of all the terms used is presented in online supplemental appendix.

How do these costing methods facilitate VBHC?

To answer RQ2, we extracted all the consequences related to the costing method as described in the papers. Here, like Etges et al,3 we were looking for how the costing information facilitated VBHC. Note that not all the studies included to address RQ1 describe facilitating VBHC or the consequences of the cost information generated. The reported consequences were grouped inductively, revealing four categories:

  1. Identification of cost drivers, in terms of cost items (eg, staff costs, material costs) or activities (eg, surgery, initial consult; n=47).
  2. Comparison of costs across patient groups, care providers or procedures (n=38).
  3. Measured cost difference or cost saving, while achieving equal or better care (n=26).
  4. Suggested or measured care path improvements (n=39).

These studies are presented in table 4. The studies reporting these facilitators used ABC (n=6), TDABC (n=28), other absorption costing methods (n=12) or direct costing (n=3).Table 4

Costing method applications, method used and consequences (ordered by year)

Activity-based costing

The six studies applying ABC justified this on the basis that it was the care provider’s existing costing method. Three of these studies measured costs for a full surgical episode21 32 34 as part of a longer care path, two measured costs for a full care path,35 36 and one measured costs of a partial care path.37 While these studies all applied ABC, the ability to facilitate VBHC differed. Jacobs et al36 measured costs for a complete care path for patients with adult spinal deformity, a complex care path spanning about 1 year. The authors compared costs across patient groups and patients, identified major cost drivers, and suggested where to concentrate cost containment. Similarly, McLaughlin et al32 34 measured costs, identified cost drivers and evaluated targeted cost containment initiatives. In one paper,32 the cost containment initiatives were informed by the cost information: activities with the highest costs were targeted for savings and a 25% reduction in total costs was achieved. In the other paper,34 they identified comorbidities and demographics that were strongly related to the total costs of patients undergoing neurosurgery, whereas Wise et al21 did not for geriatric hip fracture patients while identifying cost drivers and comparing costs across patient groups. Vanni et al35 successfully predicted about €2 million annual cost savings associated with an enhanced recovery pathway.

Time-driven ABC

The majority of the papers used to answer RQ2 involved TDABC. Significant cost drivers were identified linked to activities in a care path, and some suggested where to target improvement initiatives.4 6–8 10 33 38 39 Many of the TDABC studies were able to suggest6 8 10 19 33 38–51 or measure5 9 31 52–54 care path improvements (see table 4).

The lengths and specificities of the care path costs varied widely. Some studies were narrow in scope, calculating costs for subsections of a single care path or surgical procedure.6 8 44 48 55 Isaacson et al8 calculated costs for cleaning a single reusable piece of equipment, while others costed single surgical days,6 compared alternative surgeons,55 or anaesthesia solutions within a care path.48 Within this group, McClintock et al44 took the broadest perspective by mapping individual patient journeys.

The largest group (n=10) of TDABC studies measured costs across care paths within a single provider and for a single diagnosis.9 10 31 33 38 43 45 46 52 54 Typically, these studies compared costs between a new intervention and the ‘usual’ care,9 10 31 46 53 54 or between alternative care paths33 38 43 52 in order to measure cost savings.

Some studies were broader in scope, costing multiple care paths or treatments within one specialty,5 7 20 51 an entire department,39 40 multiple practice units47 or providers.4 Some compared ‘true costs’ calculated using TDABC across care providers within specialties or care paths,4 43 while others argued that TDABC costs were too subjective to be compared across hospitals.10 44 While most studies compared costs across care paths, some also compared costs across patient groups,19 41 42 or even individual patient journeys.38 44

Technology played a prominent role in studies aiming to reduce costs. One study was able to suggest how to use technology more efficiently,6 and some, by integrating technological investments in the calculated TDABC costs, show how technology can reduce costs.38 43 46

Conversely, studies using unspecified absorption methods56 57 did not include investments in technology, and this is surprising since absorption costing methods require indirect costs to be allocated.

Analyses enabled by activity-based and TDABC

Several of the ABC and TDABC studies compared costs calculated using traditional accounting costs6 19 20 or reimbursement amounts20 21 and found that prices do not equal costs. Some carried out quantitative analyses using cost information generated using ABC or TDABC including regression analyses to identify correlations,7 33 38 41 42 compare patient groups,20 41 42 and compare costs and outcomes across a matched patient sample.38

Two recent studies33 49 have conducted patient-level value analyses, comparing patient-reported outcomes with patient-level TDABC costs. Wise et al33 did so for rotator cuff repair surgery over a period of 1 year, while McCreary et al49 analysed ankle fractures. Both studies found costs to be unrelated to patient-reported outcome measures, highlighting the need for further research. This suggests that patient-reported outcome measures are not strongly associated with the costs of the care delivered, and that patient satisfaction may depend on other factors such as their perceived experience with healthcare professionals.

Other absorption costing methods and direct costing

Other absorption costing methods reported in the studies were labelled as microcosting (n=5), bottom-up clinical costing58 or were described but not labelled (n=6). Most were able to identify cost drivers (n=12, for details, see table 4) and some compared costs within providers. Notably, Robinson et al59 used the cost information to build and evaluate a dashboard that provides real-time feedback to surgeons during operations and monthly summaries and thereby decreases costs significantly. Some studies omitted certain cost categories such as equipment.57 Direct costing enabled cost drivers to be identified,60–62 and in some cases granular cost measurement.

Best practices

Having identified these four facilitators, we compared studies to find common practices. This is particularly useful because costing methods are not labelled consistently. For example, many studies refer to ABC as ‘bottom-up costing.’ To look beyond labels, we compared the actual methodologies used to measure costs. We found that studies that were able to facilitate VBHC used process mapping (n=33), expert input (n=17) and/or direct observations (n=24) when measuring costs. These practices overlap with TDABC best practices, but are not exclusive to TDABC, as shown in table 4.

Studies that made specific care path improvement suggestions used process mapping, and especially those involving multidisciplinary teams reported significant benefits.4 9 10 19 This approach enabled experts (doctors, care professionals, administrators) with the required knowledge and experience to reflect critically on the process,4 9 10 19 resulting in actionable suggestions. In comparison, studies that did not use process mapping tended to suggest minimising high-cost items (eg, total operating time, nursing costs) but were unable to couple these suggestions to specific activities or to chronological points in the care path. Commenting only on cost items, and not identifying chronological points, limits the ability of cost information to steer management towards where to focus process improvement initiatives.

Expert input while creating process maps or measuring costs was often cited by authors as valuable, especially for estimating preparation time or other behind-the-scenes activities that do not involve the patient but are critical to delivering care. Some studies that could not call on expert input cited this as a limitation. A few cases also evaluated the impact of costing information, for example by involving experts to evaluate a dashboard.59

Finally, some studies involved direct observations, particularly those that calculated process times to the minute or measured the costs of individual patient journeys.

Discussion

This review focused on VBHC studies that have measured or estimated costs, and on identifying which costing methods can facilitate VBHC. By assessing the consequences of the costing methods used, we were able to identify characteristics of costing methods that do facilitate VBHC.

Previous research found that TDABC can facilitate VBHC through cost containment and process improvements.3 15 We built on this by comparing value-based consequences across costing methods. While the field is young and alternatives seem limited, we have found considerable evidence that TDABC and ABC can indeed facilitate VBHC. As previously noted, TDABC is considerably easier to implement than ABC, which leads us to recommend it over ABC. We found no well-documented alternatives to TDABC or ABC in our review. However, not all the TDABC studies delivered the facilitating factors we have identified. We, therefore, emphasise the need to follow TDABC guidelines carefully and to explicitly document methods used. Several of the studies in this review simply stated that TDABC was applied, outsourced, used with incomplete costs, or used without listing exact cost rates.

The start and end points of care paths tend to be well documented by authors but are inconsistent. To view costs in relation to outcomes, as suggested by Porter,2 the total costs from start to finish of a trajectory should be included.63 In many studies, the start and end points of cost measurement windows seem somewhat arbitrary but are still labelled as full care paths. Consequently, this results in inconsistencies across studies, hindering comparisons. Encouragingly, some of the more recent studies have measured costs across a genuine full care path and future research should do the same, explicitly defining start and end points. This would enable consistent comparisons across providers. As with the ICHOM standard outcome sets produced by the International Consortium for Health Outcomes Measurement, costs could be catalogued and compared over full care paths. Indeed, in a recent expert consensus study, experts agreed on the need to focus on full care paths.63

Furthermore, we can see a trade-off in the specificity and length of the care path costed. Studies that measure costs for elements of a care path (such as a surgical operation) can provide detailed costs for that portion of the care path, but not total care costs for a patient because the remainder of the care path is not included. Some surgical studies measured costs for partial care paths, and often concluded that operating theatre time should be minimised due to high surgeon and operating theatre costs. However, this conclusion has limited relevance for the value equation13 because it does not provide cost information for an entire care path, or advice on how to circumvent surgery.

Studies that cost complete care paths appear to use less-detailed costing methods (due to the sheer length of the care path) but are able to compute total costs of a patient’s care. This enabled benchmarking across providers, as well as cost comparisons of new versus standard care, or of treatment alternatives. This allowed providers to steer towards lower-cost outcomes to maximise value. Future research should focus on measuring costs for full care paths, and on comparing costs to outcomes as demonstrated in some of the more recently published studies in our review.33 38 49

Our review highlights the need to involve medical professionals in this process, both when implementing costing methods as well as when evaluating the results. Future cost measurement studies, and hospitals looking to implement TDABC, should involve multidisciplinary teams. Studies that have involved medical professionals in the process of measuring costs and then using the findings were able to improve care paths through improvement initiatives and/or dashboards. This suggests that generating and using costing information should be viewed as a process. Future qualitative research should follow this process to better understand the mechanisms through which cost information impacts decision making, and the impact that staff involvement has on cost containment. Previous research suggests that staff involvement is critical as it builds trust in the accuracy of the data.64

Limitations and future research

We must acknowledge several limitations related to the scope, breadth and quality of the included studies. First, our search strategy will have missed studies that measure costs but do not label the study as VBHC-oriented. Not all TDABC studies make value-based claims or contributions and may therefore be overlooked in our review. In addition, not all studies explicitly discuss the impact or consequences of the costing method applied, which may impact our findings. Future qualitative research could usefully investigate TDABC implementations and evaluate whether the facilitating factors found in this review are achieved. Second, sophisticated methods such as TDABC are currently only used with predictable and/or short care paths such as orthopaedic surgery. Further research testing the feasibility and practicality of TDABC in different settings, such as emergency on-call care, or longer care paths such as fertility treatment, is warranted. Further, our findings may have limited generalisability across medical specialties as indicated in table 1. Finally, we have relied on the reporting of authors whose style and quality differs across disciplines and journals. To an extent we circumvented this limitation by looking beyond the cost measurement labels used by authors, extracting the costs included and methods used, and then categorising them using established accounting definitions. However, we cannot exclude the possibility of errors due to a lack of explicit reporting in some of the studies reviewed.

Conclusions

This systematic review reveals that cost information, at the treatment or patient level, for complete care paths does enable value-based decision making through several mechanisms. Such cost information can direct quality and process improvement initiatives alongside informing appropriate reimbursement levels. In the pursuit of VBHC, practitioners and academics are advised to apply ABC or TDABC to estimate costs, using process mapping, expert input and observations, rather than relying on pricing information.

Data availability statement

All data relevant to the study are included in the article or uploaded as online supplemental information. The data generated in this study is also available in a public, open access repository (Erasmus Data Repository, DOI: 10.25397/eur.20279883).

Ethics statements

Not applicable.

Acknowledgments

We thank W.M. Bramer, biomedical information specialist from the Erasmus MC. We also thank our fellow members of the value-based healthcare consortium, involving Amsterdam UMC, Erasmus MC, Radboudumc and Erasmus University Rotterdam, led by J.G.M. Jelsma and M.C. de Bruijne.

1670420133 Open access, Health services research