DRIVERS OF COST SYSTEM DEVELOPMENT IN HOSPITALS: RESULTS OF A SURVEY

D/2004/6482/10

Vlerick Leuven Gent Working Paper Series 2004/09

DRIVERS OF COST SYSTEM DEVELOPMENT IN HOSPITALS:

RESULTS OF A SURVEY

EDDY CARDINAELS

FILIP ROODHOOFT

Filip.Roodhooft@vlerick.be

GUSTAAF VAN HERCK

pg_0002

DRIVERS OF COST SYSTEM DEVELOPMENT IN HOSPITALS:

RESULTS OF A SURVEY

EDDY CARDINAELS

University of Tilburg

FILIP ROODHOOFT

Vlerick Leuven Gent Management School,

KU Leuven

GUSTAAF VAN HERCK

KU Leuven

Acknowledgements

The authors want to thank Greet Vandemaele for her assistance in data collection.

Contact:

Filip Roodhooft

Vlerick Leuven Gent Management School

Tel: +32 16 32 36 36

Fax: +32 16 32 85 81

Email: Filip.Roodhooft@vlerick.be

Corresponding author. Tel: + 31 13 4668231; Fax: +32 13 466 8001.

e-mail address: e.cardinaels@uvt.nl

pg_0003

ABSTRACT

While many hospitals are under pressure to become more cost efficient, new costing systems

such as Activity-based costing (ABC) may form a solution. However, the factors that may

facilitate (or inhibit) cost system changes towards ABC have not yet been disentangled in a

specific hospital context. Via a survey study of hospitals, we discovered that cost system

development in hospitals could largely be explained by hospital specific factors. Issues such

as the support of the medical parties towards cost system use, the awareness of problems with

the existing legal cost system, the way hospitals and physicians arrange reimbursements,

should be considered if hospitals refine their cost system. Conversely, ABC-adoption issues

that were found to be crucial in other industries are less important. Apparently, installing a

cost system requires a different approach in hospital settings. Especially, results suggest that

hospital management should not underestimate the interest of the physician in the process of

redesigning cost systems.

Keywords: Activity Based Costing, Organizational Change, Cost Control, Hospital context

pg_0004

INTRODUCTION

With margins on the decline, more restrictive reimbursement schemes based on

diagnostic-related groups (DRGs), increasing complexity and rising costs, the health care

sector faces a new challenge of becoming more cost efficient to survive in this changing

environment [1, 2, 3]. More developed cost systems such as Activity Based Costing (ABC),

may facilitate this strive for cost efficiency. ABC provides more detailed cost information on

the activities of the hospital, which could typically result in better cost reduction and cost

management [4, 5]. In other industries, it has proven to be successful since firms that

extensively use ABC outperform similar matched firms that do not adopt ABC, mainly

through more efficient cost control efforts [6, 7]. However, while there are different levels of

cost system design, it seems remarkable that the number of hospitals collecting cost on a more

detailed basis remains limited [2]. Relative to other industries, the health care sector still lags

behind [8]. The reason for this discrepancy has hardly ever been investigated. The main

contribution of the present study is that it provides an insight in the factors that in fact drive

(or inhibit) further cost system development in the health care sector. Via this insight,

management may better understand the crucial factors for promoting cost system

improvement in a health care environment.

As a starting point we look at ABC-adoption in other industries. We will test whether

the few existing factors known to be associated with the adoption of more accurate costing

systems in these industries, are applicable for the health care sector [9, 10, 11]. Secondly, it is

important to note that the present study takes the specific behavioral and organizational

factors of the sector into account [12]. Unlike manufacturing companies, health care providers

in many countries are for refunding purposes legally required to allocate costs in a predefined

manner e.g. Medicare Cost System in the US, [13]. Hospitals may find this legal cost system

sufficient and hence more refined costing methods such as activity-based costing may not be

considered. Important powerful coalitions [12] such as the physicians may have a stake in

whether the cost system is further developed. Thirdly, this study further recognizes that

implementing cost system refinements in hospitals typically requires progress in stages before

full adoption is achieved [2, 11].

The results of our survey, conducted in the hospital sector, show that cost system

improvement in hospitals, is largely determined by health care specific factors such as the

dissatisfaction with the legal system, the support of medical staff to cost system use, the way

the reimbursements between hospitals and their physicians are arranged etc� This seems to

pg_0005

suggest that health care management should focus on hospital specific elements in order to

facilitate ABC adoption. Factors observed in other industries have less explanatory power and

as such they may be less crucial for further promoting cost system change.

LITERATURE REVIEW

In many countries hospitals are legally required for refunding purposes to have a

predefined cost allocation scheme [13, 14]. This makes them unique to other industries where

such a legal obligation does not exist. The legal system mostly takes the form of a step-down

allocation of costs from service departments (e.g. administration, cafeteria, laundry, etc.) to

revenue generating departments such as acute care, surgery, laboratory. Sometimes cost are

further allocated down to patient-level. Often the legal system uses a large set of pre-defined

cost drivers (See U.S. Medicare cost report in Eldenburg and Kallapur [13], [15]). While such

legal systems are quite elaborated, it does not preclude management from adjusting the cost

system to make it more relevant for their internal decisions [15, 16]. Rather than immediately

installing ABC, hospitals tend to change gradually towards ABC. They often start by

adjusting their existing legal system or they may first thoroughly consider ABC [2, 9, 17]. In

that respect, hospitals seem to adhere similar implementation stages as other industries [11,

18].

Our goal is to disentangle different levels of cost system design and the drivers in a

health care setting that explain this process of changing to ABC. To our knowledge, evidence

on this matter remains very scarce. As a first step we look at general drivers of ABC-adoption

from other industries. Next, we discuss several elements from our own review of the health

care sector that may drive (or inhibit) cost system change. Finally, we provide specific control

variables for the level of cost system in a health care setting. Table 1 summarizes the drivers

we identified and their expected direction on cost system development. The next sections

further explain these issues.

Insert Table 1 About Here

pg_0006

General drivers of cost system development

There only exist a limited number of studies that identified some general drivers of

cost system improvement for firms in other industries. Below we provide more detail on those

general drivers that are expected to be relevant for a hospital setting.

Cost variability. Firms with a higher level of indirect overhead and greater

heterogeneity in the way products make use of the firm�s resources, are expected to introduce

more refined costing systems [9,11, 19]. This issue may play a role in a hospital context.

Hospitals are often known as settings with many indirect cost categories and they treat various

patients via divergent care processes that often consume overhead differently [17].

Cost importance. This issue mainly captures the way firms in other industries perceive

cost data as crucial for their decisions and their competitive position [20, 21]. Given the

current pressure on margins, this issue may especially apply to hospitals. We predict that the

stronger the importance attached to cost data, the more likely that a hospital will adapts its

cost system.

Quality link. Firms that focus on quality often link their formal quality programs with

more accurate ABC-systems [10]. Similar considerations coexist in health care. Hospitals

initiating programs to improve the quality of the care processes may be more in need of a cost

system that accurately captures the cost of these different care process [4, 17].

System State. This issue concerns the general elaboration of the IT-system within a

firm. The more elaborated and integrated the system and the more performance measures it

gathers, the easier it is to introduce ABC-systems that make use of IT-systems and their

information [22]. However, given that systems in health care often are designed to only fulfil

legal requirements [15], the culture and the resources for hospital systems to integrate

different applications and to issue performance information may not yet be well established

[2].

Perceived complexity. This issue in fact captures whether the firm�s operational

environment is perceived as complex. Arguments in other industries seem to suggest that

complex-dynamic organizations may especially benefit more from more accurate cost systems

[23, 24]. This seems true for complex organizations such as hospitals that often treat highly

complex care processes [3]. However, the perceived complexity might obstruct cost system

improvement, since the ABC problem requires very specific data from these complex

processes which may be too difficult to obtain in hospital settings [5].

pg_0007

Hospital specific elements in cost system development

Hospitals have some unique features that are typically not observed in other industries

[25]. An important contribution is that our study is one of the few to discuss the link of some

of these features with the level of cost system design in hospital settings. Below we give an

overview of these hospital specific elements.

Satisfaction legal system. As already mentioned, hospitals are legally required to use a

predefined cost allocation scheme. This unique setting allows us to test to which extent

hospitals are satisfied with this system. Due to the level of detail, satisfaction may be high

such that hospitals may not screen other cost system options [26]. Conversely, criticisms as

that the legal system would still produce unreliable cost estimates may initiate cost system

change [16].

Use legal system. This factor can be perceived as slightly different from the previous

one. While being unsatisfied about the legal system, hospitals may still consider the system

sufficient and consequently use it for their decisions. However if management questions the

usefulness of these figures [15] hospitals may be more likely to change towards refined

costing such as ABC.

Organizational support. This aspect captures the organizational support towards cost

system use. While cost innovations in other industries flow from top management support

[12], hospitals are further unique in a sense that they have to work with physicians that are

implicitly contracted without being employed for the hospital [25, 27]. As physicians are

responsible for a large part of the health care expenditures [28], their support towards cost

control in general may be important for further cost system enhancement. Besides

management and physicians, the support of the heads of various nursing departments is an

additional factor that should not be overlooked. In sum, hospitals may be further evolved on

the spectrum of cost system design when different organizational members support cost

control.

Management-physician conflict. In hospitals, physicians often perceive cost control as

very different from management. Physicians dictate that the provider-patient relationship is

quite unique and do not want to give up the freedom to deploy as much resources as needed

for the specific care of a patient [29]. This often does not stroke with ideas of central

management that needs to plan resources for the hospital as a whole [30]. It has been shown

that potential conflict between parties can arise that may hamper any innovation, such as cost

system improvements [25, 31]. Such conflict is even more likely if physicians feel that they

are controlled by central management. This is especially true if cost allocations are only used

pg_0008

for assessing (controlling) financial arrangements between physicians and hospitals [32]. Our

study assesses the level of conflict (directly by asking management to asses whether relations

with their physicians are optimal or not, and indirectly by asking the degree of control through

cost system use) as a factor that may drive or inhibit cost system change in hospitals. Cost

system improvements such as ABC are more likely when relations with physicians are less

conflicting or in other words more optimal.

Method of reimbursement. Reimbursement of health care providers (e.g. hospitals,

physicians) by health care payers (e.g. governments, insurers) typically consists of financial

flows for the operational cost of the hospital and physician labor [13, 33]. In many countries

financial flows are centrally collected by one party (mostly the hospital) who than agrees with

the other party on how to split these flows between the hospital and the physician. To this end,

several schemes exist that can either be classified as retrospective, in which the physician

receives his fee minus a payment on the basis of the own costs he incurs (physician cost

based), or as prospective in which physicians receive a fixed �percentage� of the total revenues

or financial surpluses (profit) of the hospital [34].

The reimbursement scheme may have an effect on the level of cost system design. If

they remain physician cost based (retrospective), payments are based on the indirect overhead

assigned to a specific physician [33, 34]. Management may then not be very motivated to

control costs, because physicians simply pay back most of the hospital costs. In addition

physicians may prefer a pre-defined legal cost system, as they may fear that new cost systems

give management more discretion to maximize the financial streams for the hospital [13, 35].

New ABC systems, may lead to endless debates between hospitals and physicians over the

specific assignment of overhead costs, which may hamper any cost system change [14].

Conversely under prospective systems, payments are at least not physician cost based.

Furthermore, if payment is based on surplus (profit) rather than on total revenues this may

create some incentives for cost control and as such there may be a need for ABC [36].

Specific control variables

Prior work suggests a positive relation between firm size and the level of ABC-

adoption [9, 10, 11] did not find such an effect. Evidence in the health care sector suggests

that larger hospitals in terms of bed size more extensively use their cost system [2]. We

therefore take �Bedsize� as a first potential control variable of the level of cost system

development. As a second control variable we check whether hospitals are involved in a

pg_0009

merger. Those hospitals that struggle for survival are often restructuring their operations via

mergers and therefore limited resources are not spent on improving the cost systems [2].

Mergers take up most of the time and cost system improvements are probably postponed until

the merger is completed.

RESEARCH METHOD

Research Sample

The survey was conducted on a sample of hospitals, located in Flemish part of

Belgium. Similar to most other countries, all hospitals in our sample are required to issue a

legal cost report based on an elaborated set of drivers in a step-down allocation scheme from

service to revenue generating departments. In addition, these hospitals also agree on various

reimbursement schemes with their physicians. A total of 120 questionnaires were issued to

either general hospitals, academic hospitals, psychiatric hospitals or specialized hospitals. The

survey administered questions to identify the stage of cost system development and the

hospital specific and general drivers that are possibly linked with the level of cost system

design (sections 3.2 and 3.3 give more detail about the survey items). The survey was either

addressed to the chief executive officer of the hospital facilities or the chief of the

administration and financial department. These respondents are most likely to be informed

about the design and the use of cost systems in their hospital.

Of the 120 questionnaires, we received 50 valid responses. This corresponds to a

response rate of about 42%. Of the 50 valid replies, 48% came from general private hospitals,

10% from general public hospitals, 38% from psychiatric facilities and the remaining 4%

from either academic or specialized private hospitals. It is important to note that the sample�s

distribution is not significantly different from the distribution within the total population of

120 Flemish hospitals (Chi-square: 2.3; p = 0.13). In terms of size our sample counted 20%

small facilities with less then 200 beds, 56% intermediate-sized hospitals with 200 to 499

beds and 24% large hospitals with over 500 beds.

Dependent variable

The primary dependent variable for our study is the stage of cost system development.

Via our survey study we were able to identify three possible levels of cost system design. A

first group of hospitals only installed the legal system. A second group of hospitals is in the

process of changing their cost system. Either they started with small adjustments to their legal

pg_0010

system by introducing more specific drivers and cost objects (e.g. patient-levels, DRG-levels)

or they were in the process of considering ABC [2, 9]. This group may be situated on a sort of

�intermediate level� in the process of change towards more refined costing systems. The last

group is on a more advanced level of cost system refinement. They actually indicated to be

experimenting with ABC (Cfr. adoption phase; [11]) and as a result of this exercise they

developed an adapted cost system. Table 2 shows how the sample of 50 hospitals is

distributed across these three possible development stages of cost system design. One should

further note that hospitals in phase 1 are somehow distinct from the two other groups. Unlike

hospitals in phase 2 and 3, these hospitals do nothing in terms of cost system refinement. In

the result section, we report an additional model based on this dichotomy.

Insert Table 2 About Here

Independent variables

The general drivers and most of the hospital specific elements, except for the type of

reimbursement scheme, were measured via multiple (e.g. two or more) items that were in fact

based on our arguments of the literature review. Appendix A displays the set of items issued.

Respondents indicated the relevance for each item on a five-point Likert-scale (1= strongly

disagree; 5= strongly agree). A first set contains items for the general drivers such as cost

variability, cost importance, quality link, system state and perceived complexity. The next set

focuses on the remaining hospital specific issues such as organizational support, satisfaction

with and the use of the legal system and the level of conflict between management and

physicians. We preferred multiple items because they capture more of a construct than single

items [1, 37]. However to test whether our items actually capture the presumed construct,

factor analyses were performed on both the sets of general drivers and hospital specific

factors. The results of these factor analyses are displayed in panel A of table 3. Results show

that the derived factors correspond closely to the constructs of the literature review, save for a

few exceptions that will be discussed below.

Regarding the general drivers, it is important to note that the construct cost variability

and cost importance form one factor �Cost_var�. Apparently greater cost variability is a

synonym for more importance attached to cost data. All items of the second factor

�Syst_state� indeed relate to the state of IT-systems in the hospital. The third factor

�Complexity� forms the construct for the perceived complexity of the hospital processes and

pg_0011

the cost allocation. Finally, we mention that our last factor does only partially captures our

construct for the link of the cost system with quality. It only loads high on the quality item F

(Table A1 in Appendix A). However, this last factor has also high loadings on item G

measuring the extent to which systems generate various performance measures. We label this

factor �Perf_link� as the degree of focus on performance measures in a hospital. Shields [12]

suggests that this issue may indeed be relevant if ABC adoptions want to succeed. Analysis on

the hospital specific items resulted in four factors with main items that indeed correspond to

the presumed construct. Only the second factor related to organizational support does not load

high on management support (Item L), suggesting that the views of management on cost

control are divergent from the views of the medical staff. We label this factor �supp_med� as

the support of medical parties towards cost control. The other factors are labeled as

�sat_legal�, �use_legal� and �conflict� according to their construct.

Similar as to Krumwiede [11, p. 249-250] we want use the factors as independent

variables for explaining the level of cost system design (section 3.1) To this end, we

calculated for each hospital a composite score for the derived factors. A composite factor

score is an aggregated score of responses giving the most weight to items that load high on

that specific factor. On average, they have a mean of zero and a standard deviation of 1 and

correlations between factors approximate to zero. Alpha levels on the main items indicate that

factors appear to be reliable and reasonably valid.

Finally, the remaining three independent variables, that is the hospital specific factor

for the type of reimbursement and our two control variables, were measured directly via a

single question. These variables are summarized in panel B of table 3. The variable

�Reimbursement� was based on a dummy. It is derived from the question in which

respondents indicated whether the reimbursement scheme was based on physician specific

cost elements such as actual cost or actual cost plus mark-up (Reimburse= retrospective) or on

a fixed percentage of revenues or hospital surpluses (Reimburse= prospective). Next, the

number of beds for each hospital facility represented our first control variable �Size� while

our second control variable �Merger� is a zero vs. one variable (dummy) depending on

whether or not a hospital indicated to be highly involved in restructuring its operations (e.g.

merger).

Insert Table 3 About Here

pg_0012

EMPIRICAL FINDINGS

We in fact performed two analyses. The first section uses the three levels of Table 2 as

the dependent variable. In this way we can derive the factors that significantly differentiate

between the various stages of cost system design, that is the drivers of cost system refinement.

In the next section we study the dichotomy of hospitals that do not perform any cost system

refinement (minimum level) versus all others that change. This analysis should shed light on

the first initiators of cost system change.

Drivers of cost system development

Because of the specific order in the level of cost system design, an ordered logistic

regression is actually the most appropriate method for this analysis. Hospitals on an advanced

level (level 3) are further on the spectrum of cost system design than hospitals in the process

of change (level 2) or those that only have a legal system (level 1). Model 1 in Panel B of

Table 4 reports the results of this regression.

When studying the general drivers, we only observe a significant positive effect of the

variable �cost_var�. Apparently hospitals that perceive high variability in costs and that attach

high importance to cost in general are more likely to adjust their cost system in the direction

of ABC. Summary statistics in Panel A of Table 4 show that especially the hospitals that have

changed their system as a result of ABC-adoption (advanced), seem to find this issue much

more important (higher factor score) than those hospitals that are in the process of changing or

that only have a legal system. The state of IT-systems, the perceived complexity and the link

with performance (including quality) do not drive or inhibit cost system change in a hospital

setting.

Regarding the hospital specific elements, we observe more significant effects. First of

all, �satisfaction with the legal system� is significant and has a negative sign (model 1 in panel

B). From panel A we can argue that hospitals that are less satisfied with the legal system are

more likely to change or to install ABC (level 2 and 3) compared to their counterparts that

only use a legal system (level 1). Although the system is quite elaborated, some Belgian

hospitals seem to be unsatisfied as a result of perceived shortcomings to the legal system [15,

16] and consequently these hospitals are more likely to improve their cost system.

Panel A and Model 1 in Panel B further suggest that high support of the medical team

towards cost control (Supp_med) is a factor that significantly differentiates among the

different stages of cost system design. Unlike in other firms where cost system changes go

pg_0013

through top management [12] our results point out that physicians, medical boards and heads

of nursing departments seem to be powerful coalitions that may further stimulate changes

towards ABC in hospital settings.

As suggested in our literature review, the reimbursement scheme is significant.

Evidently, when reimbursements are physician cost based (retrospective) rather then

prospective (e.g. fixed percentage of revenues or surplus), hospitals are less likely to change

to ABC. Panel A indeed shows that none of the respondents in phase 3 had a reimbursement

scheme based on physician costs (retrospective), while there are still a large number of users

of retrospective schemes in phase 2 (45,8%) and phase 1 (55,0%). Under retrospective

systems, physicians may fear that hospitals will use cost system changes to alter the cost-

based amount physicians have to refund [36]. At least prospective schemes are not based on

cost allocations and if they further use a fixed percent of hospital surpluses (instead of

revenues), they may stimulate a need for better cost control in order to increase the hospital

surplus.

Our two remaining hospital specific factors �conflict management-physician� and �use

legal system� do not seem to differentiate among the different development stages. However,

not only arguments of our literature review but also evidence from correlation tests

allude to

a possible link of the reimbursement scheme with these two variables. When reimbursements

are based on cost allocations (retrospective), there is more conflict between management and

physicians probably resulting from debates over which cost to include in the analysis.

Secondly, a likely explanation why retrospective systems may be linked to higher use of the

legal system is that physicians may prefer (or force) the legal system for cost reimbursements.

Unlike with new cost allocations where management may change allocation bases to

maximize financial streams for the hospital [13], the legal system uses at least pre-defined

cost allocation bases, so that hospital management has less discretion to maximize cost

reimbursements emanating from the physician.

Due to these interactions, possible effects of �use_legal� and �conflict� may not be

observed in model 1. We therefore ran model 2 in which reimbursement was left out the

regression. Results show that �conflict� and �use_legal� become significant. In sum this hints

that cost system changes are more likely when there is little conflict between management and

Correlations of conflict and reimbursement (r: -0.367; p: .009) suggest that relations with physicians are less

optimal when reimbursements are retrospective. In addition legal systems are also used more when

reimbursement is physician cost based, though this correlation is weaker (r: 0.262; p: .066).

pg_0014

physicians and when legal systems are considered as less useful for decision-making, which

may in turn be driven by the type of reimbursement scheme.

Finally, our variables do not load significantly in both our two models. Apparently the

hospital�s size and its involvement in mergers do not differentiate between the different

development stages that our survey identified

Insert Table 4 About Here

Minimum level vs. the changers

To single out the first initiators of change, we perform a binary logistic regression of

those hospitals that do not change (Minimum: level 1) vs. all others that change (level 2 and 3

are taken together). Results are reported in model 3 and 4 of Table 4 and are similar to the

models reported earlier, except for the fact that �Cost_var� is not significant anymore. The

models suggest that the hospital specific factors such as the satisfaction with the legal system,

the support of medical parties and the method of reimbursement (and climate if

reimbursement is left out of the analysis) serve as the first initiators of change. �Cost_var� a

general driver becomes only important in later stages if we recognize the difference in

intermediate level and advanced level (models 1 and 2), but not in the current analysis.

Summary statistics indeed confirm that this general driver especially matters at the more

advanced level of cost system design.

Implications of the results

Hospitals tend to follow similar stages of cost system refinement as other industries.

Our results however suggest that hospitals should stimulate health care specific issues rather

than the general drivers of other industries. Only the level of cost variability and cost

importance as a general driver is important only at more advanced levels of ABC adoption.

Hospital specific issues in fact serve as initiators of change towards ABC. Especially the

support of the medical staff should be considered if hospitals refine their cost system. Other

measures such as the awareness of limitations of the legal system can further initiate cost

system change. Of special interest is that management may need to revise the method of

reimbursements between hospitals and physicians in order to ease ABC-adoption. If

Other measures for size, e.g. the number of full-time employees, were also not significant.

pg_0015

reimbursements remain physician cost based ABC adoption is difficult; cost system change

may then further be precluded because of more conflicts and greater use of the legal system.

DISCUSSION

As hospitals� income is under pressure as a result of rising health care costs and more

restrictive budget constraints, hospitals are looking for options to become more cost efficient.

For assisting their strive for cost efficiency, health care organizations may want to adopt more

refined costing techniques, such as activity based costing (ABC) as they have proven to be

successful in other industries [6]. However the factors that facilitate (or inhibit) this change

towards ABC have not yet been investigated in hospital settings. Via a survey we single out

factors that explain further cost system development in a health care context. First of all, the

survey shows that similar to other industries cost system change in hospitals gradually

happens in different stages. However and more importantly, results indicate that the general

drivers of ABC adoption from other industries are less crucial for promoting cost system

change in hospitals. Apparently, typical features of the health care sector such as the

satisfaction with and the use of the existing legal system, the support of the medical team, the

level of conflict with and the way in which physicians are reimbursed seem to explain

variations in cost system development among hospitals.

Hospitals are quite unique settings in a sense that they have to work with highly

autonomous groups of physicians [25, 27]. While cost system changes normally flow from top

management [12], our results suggest that in hospitals physicians and other medical parties are

apparently powerful coalitions when it comes to redesigning cost systems. Not only the

support of the medical team towards cost system change, but also a minimal level of conflict

with the physician, make cost system change towards ABC more likely. The way hospitals

arrange their reimbursement with the physicians may also require reassessment. If refunds

depend on cost allocations, there may be endless debates over which cost to include in the

analysis. Furthermore, physicians are not likely to go along with cost system changes as new

cost systems such as ABC may give hospitals more discretion to maximize the cost

reimbursement streams from the physician. Conversely changing to ABC is easier if

reimbursements are not physician cost based. In sum, it is important for hospitals to consider

the stakes of the physician and their support towards cost systems in the process of cost

system refinement.

pg_0016

The fact that specific issues of the sector are more crucial for promoting cost system

change may explain why hospitals typically lag behind other firms. Installing ABC apparently

requires a different approach in hospitals. For example, the change of attitude of the

physician, installing new reimbursement schemes may require time that can slow down the

process of changing towards ABC. We however do not depict factors of other industries as

not important. Hospital specific factors may be the first steps of cost system change, while

general drivers may become highly important in later stages (e.g. this applied to a certain

extent for the general driver cost variability). The quality of IT-systems, top management

support, the link with performance and quality measures, the perceived complexity may all be

crucial factors in the process of ABC to grow to a fully operational system. Unfortunately, we

only had a limited number of hospitals that adapted their cost system via ABC. Therefore, it is

difficult to recognize further divisions in the type and the level of ABC-systems within this

group. We however leave this fascinating conjecture for future research.

pg_0017

APPENDIX A

Insert Table A1 about here

pg_0018

REFERENCES

Madorr�n Garcia C, Val Pardo I. Strategies and performance in hospitals. Health Policy

2003; Forthcoming.

Hill N, Johns, E. Adoption of costing systems by U.S. Hospitals. Journal of Health Care

Management 1994; 39 (4): 521-537.

Udpa S. Activity-based costing for hospitals. Health Care Management Review 1996; 21

(3): 83-96.

Devine K, O�clock P, Lyons D. Health-care financial management in a changing

environment. Journal of Business Research 2000; 48: 183-191.

Lawson R. Development of an HMO cost management system. Research in Healthcare

Financial Management 2003; 8 (1): 31-41.

Ittner C, Lanen W, Larcker D. The association between activity-based costing and

manufacturing performance. Journal of Accounting Research 2002; 40 (3): 711-726.

Kennedy T, Affleck-Graves J. The impact of Activity-Based Costing techniques on firm

performance. Journal of Management Accounting Research 2001, 13: 19-45.

Grazier K. The future of management care. Journal of Healthcare Management 1999; 44

(6): 423-426.

Bj�rnenak T. Diffusion and accounting: the case of ABC in Norway. Management

Accounting Research 1997; 8: 3-17.

10.

Innes J, Mitchell F, Sinclair D. Activity-based costing in the U.K.�s largest companies: a

comparison of 1994 and 1999 survey results. 2000; 11: 349-362.

11.

Krumwiede K. The implementation stages of activity-based costing and the impact of

contextual and organizational factors. Journal of Management Accounting Research.

1998; 10: 239-277.

12.

Shields M. An empirical Analysis of firms� implementation experiences with activity-

based costing. Journal of Management Accounting Research. 1995; 7: 148-167.

pg_0019

13.

Eldenburg L, Kallapur S. Changes in hospital service mix and cost allocations in response

to changes in Medicare reimbursement schemes. Journal of Accounting and Economics

1997; 23: 31-51.

14.

Peden A, Baker J. Allocating physicians� overhead costs to services: an econometric/

accounting-activity-based-approach. Journal of Health Care Finance 2002; 29 (1): 57-75.

15.

Jackson T. Using computerised patient-level costing data for setting DRG weights: the

Victorian (Australia) cost weight studies. Health Policy 2001; 56: 149-163.

16.

Kerschner M, Loper E. Enhancing cost accounting efforts by improving existing

information sources. Topics in Health Care Financing 1987; 13 (4): 10-19.

17.

Ridderstolpe L, Johansson A, Skau T, Rutberg H, Ahlfeldt H. Clinical process analysis

and activity based costing. Journal of Medical Systems 2002; 26: 309-322.

18.

Cooper R, Zmud, R. Information technology implementation research: A technological

diffusion approach. Management Science 1990; 36 (2): 123-139.

19.

Gupta M, King R. An experimental investigation of the effect of cost information and

feedback on product cost decisions. Contemporary Accounting Research 1997, 14 (1): 99-

127.

20.

Estrin T, Kantor J, Albers D. Is ABC suitable for your company? Management

Accounting 1994 (April): 40-45.

21.

Swenson D.

The benefits of activity-based cost management to the manufacturing

industry. Journal of Management Accounting Research 1995; 7: 167-180.

22.

Cooper R. The rise of Activity-based costing: when do I need an activity-based costing

system. The Journal of Cost Management 1988, Fall: 41-48.

23.

Anderson S. A framework for assessing cost management system changes: the case of

activity-based costing systems at General Motors, 1986-1993. Journal of Management

Accounting Research 1995; 7: 1-51.

pg_0020

24.

Gul F, Chia M. The effects of management accounting systems and environmental

uncertainty and decentralization on managerial performance: A test of three-way

interaction. Accounting, Organization and Society 1994; 19 (4/5): 413-426.

25.

Eldenburg L. The use of information in total cost management. The Accounting Review

1994; 69 (1): 96-121.

26.

Innes J, Mitchell F. A survey of activity-based costing in the U.K.�s largest companies.

Management Accounting Research 1995: 137-153.

27.

Eeckloo K, Van Herck G, Van Hulle C, Vleugels A. From corporate governance to

hospital governance: Authority, transparency and accountability of Belgian non-profit

hospitals� board and management. Health Policy 2004, Forthcoming.

28.

Berndtson K. Management and Physicians come head to head over cost control. Health

Care Financial Management 1986; 40 (9): 22-26.

29.

Broyles R, Reilly B. Physicians, patients, and administrators: a realignment of

relationships. Journal of Healthcare Management 1998: 33 (1): 5-14.

30.

Wood K, Matthews G. Overcoming the physician group-hospital culture gap. Healthcare

Financial Management 1997; 51 (3): 69-70.

31.

Mintzberg H. Toward healthier hospitals. Health Care Management Review 1997; 22 (4):

9-18.

32.

Bettner M, Collins F. Physicians and administrators: inducing collaboration. Hospital &

Health Services Administration 1987; 32 (2): 151-160.

33.

Leone A. The relation between efficient risk-sharing arrangements and firm

characteristics: Evidence from the management care industry. Journal of Management

Accounting Research 2002; 14: 99-117.

34.

Jegers M, Kesteloot K, De Graeve D, Gilles W. A typology for provider payment systems

in health care. Health Policy 2002; 60: 255-273.

35.

Zimmerman J. Accounting for Decision Making and Control (Irwin, Chicago, IL) 1995.

pg_0021

36.

Greene J, Metwalli A. The impact of activity based cost accounting on health care capital

investment decisions. Journal of Health Care Finance 2001; 28 (2): 50-64.

37.

Foster G, Swenson D. Measuring the success of activity-based management and its

determinants. Journal of Management Accounting Research 1997; 9: 107-139.

pg_0022

TABLE 1

Relevant issues in cost system development

General drivers

Hospital specific issues

Control variables

Cost variability (+)

Cost importance (+)

Quality link (+)

System state (+/-)

Perceived complexity (+/-)

Satisfaction legal system (-)

Use legal system (-)

Organizational support (+)

Management-physician conflict

(+ if less conflict)

Reimbursement (retrospective, -)

Hospital size (+)

Involved in merger (-)

pg_0023

TABLE 2

The different phases of cost system development identified by the survey

Phases of cost system development

Numbe

r of

Hospit

als

Percenta

1. Minimum: Only the legal system

40%

2. Intermediate: Process of changing the cost system

48%

3. Advanced: Adapted cost system as result of ABC adoption

12%

Total

100%

pg_0024

TABLE 3

Definitions of the independent variables

PANEL A: Independent variables as a result of a factor analysis

Variable

Definition and main items

(item info in appendix A)

Variance

Explained

Reliability

(Alpha)

range composite

factor score

Factor analysis on the general drivers, 4 factors extracted:

Cost_Var The importance of cost data

and the variability of costs

21,96%

0.7433

-2,21 to 1,32

(items A, B, C, D)

Syst_State The quality of information

15,58%

0.6693

-2,11 to 2,31

Systems

(items G, H, I)

Complex The perceived complexity of

14,77%

0.5217

-2,47 to 1,81

the hospital environment

(items J, K)

Perf_Link Extent to which performance

measures are used in hospital

13,58%

0.6382

-1,85 to 2,42

(items F and G)

Factor analysis on the hospital specific elements, 4 factors extracted:

Sat_Legal Satisfaction with legal system

and its perceived accuracy

23,99%

0.8976

-1,62 to 3,34

(items P, Q, R)

Supp_Med The importance that medical

20,52%

0.8418

-2,00 to 2,28

Parties attach to cost system

(items M, N, O)

Use_Legal The extent to which legal

14,80%

0.5185

-2,13 to 2,49

System is used for decisions

(items S, T, U)

Conflict

Level of management-

physician conflict

12,30%

0.6313

-2,62 to 2,10

(items V, W inverted)

PANEL B: Independent variables based on a single question

Variables

Definition

Size (contol)

The number of beds of a hospital facility

Merger(control)

Dummy for whether a hospital is involved in restructuring

operations (0 for low involvement; 1 otherwise)

Reimburse (hospital) Dummy for reimbursement scheme; 0 for prospective; 1 if it is physician cost

based (retrospective)

Factors extracted using the principle component analysis (rotated solution; Eigenvalues all > 1)

Alpha based on the main items between brackets (Cfr. items with the highest loadings for that factor)

Higher scores actually represent a more optimal relation and hence a lower level of conflict

pg_0025

TABLE 4

Summary statistics and regression results

Panel A: Average statistics of the variables (factor scores) for each cost system phase

Phase 1

Phase 2

Phase 3

Minimum intermediate advanced

General

Cost_Var

-0,28

-0,01

0,96

Syst_state

0,18

-0,20

0,19

Complex

-0,03

0,12

-0,36

Perf_link

-0,41

0,33

0,04

Hospital

Sat_Legal

0,55

-0,41

-0,21

Supp_med

-0,49

0,23

0,74

Use_legal

0,17

-0,03

-0,46

Conflict

-0,23

0,07

0,49

Reimburse (%retrospective)

55,0% 45,8%

0,0%

Control

Size (Average No. Beds)

331

426

402

Restruct (% highly involved)

30,0%

58,3%

33,3%

Note that the conflict variable uses the inverted score of item W. A higher score means less conflict as

the relation with the physician is more optimal and costs are less used for financial control purposes.

Panel B: Regression results

Ordered logistic regression

Three development stages

Binary logistic regression

Minimum level versus changers

Model 1

Model 2

Model 3

Model 4

Variable

Estimate (sign.) Estimate (sign.)

Estimate

(sign.)

Estimate (sign.)

Coeff_1

0.249 (.633) -0.382 (.438) 0.854 (.440) -0.618 (.391)

Coeff_2

2.875 (.001)

***

2.016 (.001)

***

General

Cost_Var

0.588 (.019)

0.481 (.038)

1.023 (.102) 0.468 (.265)

Syst_state

-0.082 (.693) -0.024 (.904) -0.609 (.168) -0.263 (.426)

Complex

-0.210 (.304) -0.226 (.176) -0.128 (.674) -0.122 (.634)

Perf_Link

0.208 (.365) 0.116 (.592) 0.713 (.207) 0.332 (.388)

Hospital

Sat_Legal

-0.750 (.002)

***

-0.630 (.002)

***

-1.619 (.009)

***

-1.135 (.005)

***

Supp_Med

0.738 (.003)

***

0.697 (.003)

***

1.108 (.038)

0.902 (.046)

Use_Legal

-0.287 (.193) -0.423 (.044)

-0.171 (.669) -0.445 (.186)

Conflict

0.261 (.266) 0.474 (.030)

0.582 (.178) 0.693 (.076)

[Reimburse=1] -1.183 (.012)

-1.863 (.059)

Control

Size

4.2e-04 (.634) 4.2e-04 (.623) 1.9e-03 (.186) 9.5e-04 (.420)

[Restruct=1]

0.297

(.493) 0.271

(.512) 0.621

(.415) 0.952

(.143)

Chi-square model 41.71 (.001)

***

35.10 (.001)

***

40.47 (.001)

***

35.46 (.001)

***

Pseudo R-square 0.566

0.504

0.555

0.508

dependent: Y=1 (minimum), Y=2 (intermediate), Y=3 (Advanced)

dependent: Y=0 (Only a legal system, minimum); Y=1 (Changers=intermediate & advanced)

*,**,***, significant at respectively 10%, 5%, 1% level

pg_0026

TABLE A1

Item list (used in factor analyses) and summary statistics per item

Percentages

Items

1 2 3 4 5 mean S.D.

General drivers in other industries

Cost variability

A. Certain care processes (DRG�s), patients

require more costs than others

2% 2% 22% 20% 54% 4,22 1,00

B. The indirect costs constitute a larger part of

total costs

0% 10% 24% 34% 32% 3,88 0,98

Cost importance

C. Cost information is important for staying

competitive as a hospital

2% 6% 12% 27% 53% 4,24 1,01

D. Accurate cost data is crucial for our hospital

0% 0% 4% 34% 62% 4,58 0,57

Quality link

E. Total Quality Management of our health

care processes is a very important issue

0% 2% 18% 31% 49% 4,27 0,83

F. Our personal is rewarded for improving

the quality of service to the customer

14% 45% 31% 6% 4% 2,41 0,94

System State

G. Cost systems are linked to a spectrum

of different performance measures

6% 33% 27% 29% 4% 2,92 1,02

H. The various IT systems (electronic patient

files, inventory) are strongly integrated

16% 31% 29% 20% 4% 2,65 1,09

I. It is difficult to use our systems for defining

standard activities at the patient level

2% 18% 27% 39% 12% 3,38 1,03

Perceived complexity

J. Care process in our hospital are highly complex

0% 4% 25% 45% 24% 3,89 0,81

K. For our specific hospital it is complex to

allocate cost in an accurate manner

8% 36% 28% 26% 2% 2,78 1,00

2. Organizational and behavioral items within health care

Organizational support

L. The board of directors strongly supports

cost allocation (top management)

7% 7% 35% 39% 13% 3,46 1,03

M. The medical board strongly supports cost

system use (physician)

21% 19% 47% 12% 2% 2,56 1,03

N. The physicians strongly favor the use of

cost systems (physician)

26% 19% 42% 12% 2% 2,47 1,08

O. Heads of various nursing departments

support cost control (nursing)

23% 21% 46% 10% 0% 2,44 0,97

Satisfaction legal system

P. We are satisfied with the legal costing system

14% 37% 31% 16% 2% 2,55 0,99

Q. Cost drivers of the legal system allocate cost in

a logical manner

12% 45% 31% 10% 2% 2,45 0,90

R. Cost calculated under the legal system quite

accurately reflect the true cost

14% 51% 24% 10% 2% 2,35 0,91

Use legal system

S. The legal system is easy to use

6% 24% 16% 39% 14% 3,34 1,17

T. The legal system is not optimal but it satisfies

our decision needs

10% 33% 33% 16% 8% 2,78 1,08

U. The legal system is often used in our decisions

20% 25% 24% 24% 8% 2,75 1,25

Conflict management-physician

V. Our relationship with our team of physicians

can be described as optimal

4% 18% 22% 49% 8% 3,39 1,00

W. Cost allocation is only a necessity in

managing financial relations with our

physicians

37% 35% 24% 2% 2% 1,96 0,94