How to Measure Return on Investment - Semantic Scholar...M. Imhoff: How to Measure Return on Investment 7 As it is impossible to cover all perspectives of ROI analysis in health care

M. Imhoff: How to Measure Return on Investment 1

How to Measure Return on Investment

Michael Imhoff

Dr. med. Michael Imhoff

Chirurgische Klinik

Städtische Kliniken Dortmund

Beurhausstrasse 40

D-44137 Dortmund

Germany

Tel: +49-231-50-20021

Fax: +49-231-50-20081

email: [email protected]

Abstract

As a result of the increasing pressure on health care expenditures calculations of the return on

investment (ROI) have been attempted in assessing the utility of medical equipment purchases.

The measurement of ROI requires an in-depth analysis of equipment cost, cost structure, and

medical processes. ROI analysis in health care is more complicated and error-prone than in most

other industries. Results may dramatically vary between different countries even for the same

investment. This article will outline a practical approach to ROI analysis in health care.

Keywords

Amortization - medical equipment - medical information technology - reimbursement - cost-ef-

fectiveness analysis


Introduction

Under the increasing pressure on health care expenditure hospital administrators and health care

professionals have been forced to focus their decision-making more and more on cost and cost

effectiveness of care. Just one or two decades ago most investments into medical equipment,

facilities, or staff were seen from a medical perspective, such as providing for better treatment,

advanced medicine, or satisfying customer, i.e., patient, requirements. Today investments must

prove their pay back also in monetary terms. It has become more important not only to measure

but also to forecast the amortization of investments into health care [1, 2, 3, 4, 5].

Therefore, it comes to no surprise that since 1970 about 400 articles with reference to return on

investment (ROI) are listed in the MEDLINE database [6]. Despite the sheer number of articles

the clinician remains dissatisfied, as most of these articles deal with pharmaceutical develop-

ment and veterinary medicine (food production). Only very few articles provide information

about the theory and practice of ROI analysis in the clinical field. Therefore, this article tries to

cover at least some clinically important aspects of this topic.

Webster's dictionary defines return on investment in the following terms: “return is the profit

derived from investment. Likewise return on investment is the profit that the investment

yields.”[7]

The terms and definitions of “return on investment” and “cost effectiveness” have their origins

in the economic literature rather than in medical science [8]. Therefore, most of our knowledge

and methods for measuring ROI do come from non-medical applications. If we compare the

health care industry with other productive and service industries there are many aspects that

may significantly affect ROI analysis. For instance, in other industries


• increasing overall revenue is one important aspect of ROI,

• hard data about the production process, costs, and retail prices are available,

• easily quantifiable products are produced, and

• the rules of a free market can be applied.

In health care the situation is more complicated:

• There are many external restraints and most countries do not have a free health care market.

• There are certain quality requirements, for instance, to avoid litigation for malpractice, or to

satisfy patients' requirements. But this is in some way comparable to other industries like avi-

ation [3].

• The consumer, i.e. the patient, does not always have to pay directly for the service he makes

use of. Therefore, the medical customer may have different appreciation of health care serv-

ices than of other, non-medical products.

• The investor may not always be the party receiving the ROI. For instance, in many regions

and countries capital investments into hospitals may be supported by state or federal funds,

while the operating costs of the hospital are covered by the hospital's budget. In this scenario

the investor, e.g., the state, may not have a strong incentive to invest more to reduce the op-

erating cost of the hospital, as this may only benefit the hospital's but not the state budget.

• Health care organizations, e.g., hospitals or independent delivery networks, may not be al-

lowed to keep their profits, as is the case in several European countries. Therefore, they do

not have a strong incentive to make investments that may in the long run increase their prof-

itability.


Another very important aspect in the analysis of ROI are the different reimbursement methods

that generate the hospital’s revenue [9]. Even within one health care organization, but always

across the different health care systems entirely different reimbursement methods may be en-

countered, as there are:

• Fee for service,

• case base and capitation,

• disease related groups (DRG),

• per diem,

• and combinations thereof.

It is obvious that these different reimbursement methods will lead to an entirely different finan-

cial bottom-line for the same investment and the same change in the medical process. An invest-

ment that may yield a significant ROI in one health care environment may result in a dramatic

loss of money in another.

The scope of ROI in health care

In other industries the scope of ROI most often is very clear. It is typically only analysed from

the perspective of the investor, which in most cases is also the company benefiting from this

investment. Normally this perspective is the only applicable one.


In health care there may be many different perspectives to assess ROI. We need also to keep in

mind that investor and beneficiary are not always identical. Therefore, we need to define which

perspective we address before we engage in a ROI analysis. Different perspectives in health care

may include

• the individual care provider, e.g., the general practitioner,

• the health care organization, e.g., a hospital,

• a health management organization (HMO),

• an entire health care system, i.e., the public purse,

• the employer of the patient, or

• the individual patient.

Two examples may show these different perspectives.

In the first scenario a radiology practice privately invests into a new MRT. This new device will

allow additional MRI studies and by increasing the number of studies will increase the revenue

of the radiology practice. In a fee for service environment this may yield a significant ROI for

the respective radiology practice. On the other hand, the payer or health care system will have

to carry additional expenses.

In the second scenario a newly developed clinical pathway system significantly shortens the

length of stay after laparoscopic cholecystectomy from five to two days. In a case base, capita-

tion, or DRG environment this may lead to a ROI for the respective hospital, while being cost

neutral for the payer. If the reimbursement is done on a per diem basis only the payer may ben-

efit, while the hospital runs into bankruptcy.


As it is impossible to cover all perspectives of ROI analysis in health care in one article, this

article will focus on the perspective of the health care provider, the hospital. We will assume a

highly regulated health care setting with reimbursement on a per case basis or DRGs. We will

outline the general layout of ROI analysis and take a closer look at the different factors that may

influence ROI. One hypothetical example will be used to illustrate the complexity and intrica-

cies of ROI analysis in health care in different settings.

How to generate ROI in health care

As mentioned before there are stark differences between health care and other industries in how

ROI can be generated [8, 9]. As in other industries also in health care there are two basic ap-

proaches to ROI, namely increasing the revenue of the operation or reducing the cost per pro-

duction item. The differences are how likely each of these approaches can be followed in health

care compared to other industries.

Basically, there are three ways to increase the revenue from the operation into which the invest-

ment was done.

First, the investment can be used to expand the existing market or to find new markets. This ap-

proach is extremely unlikely in regulated health care environments. On the other hand, it is an

opportunity in non regulated settings, such as aesthetic surgery.


Second, the investment aims at taking a market share from competitors. The likelihood of suc-

cess for this approach depends on the structure of the health care system. While in many highly

regulated European health care systems even the prospective number of patients per quarter or

year that a hospital is allowed to treat may be regulated, and therefore, increasing the market

share for an individual hospital may be impossible, in a more competitive health care setting,

such as US HMOs, this approach may be successful.

Third, the investment will be used to improve the quality of the product offered to the customer.

Consecutively the customer will be charged a higher price for the better product. This approach

will typically not work in any given health care environment.

Today, in health care the emphasis is on cost reduction [2, 10]. This theme is constantly reiter-

ated in all discussions about health care expenditure.

The basic goal is to reduce the cost per service item, while a service item could be a specific

medical procedure, e.g., an MRI study, one specific disease episode, the complete treatment for

one DRG, or even the entire health coverage for a population. In principle, there are three ap-

proaches to reduce cost:

• Reduction of resource consumption, i.e., the direct cost, for an individual service item. This

could, for instance, be achieved by a medical device that has lower operating costs.

• Increase of the number of procedures done with the same resource, i.e., reduction of the fixed

cost per service. An example could be extended operating hours for expensive imaging

equipment. Of course, this reduction of fixed costs needs always be calculated against the

absolutely increased direct costs.


• Reduction of complications for the delivered service item. This approach to reduce cost is

highly specific for health care. Typically, complications are excessively expensive, as they

require additional treatment, increase the length of stay, and may result in compensation for

malpractice. Therefore, even a small reduction in the complication rate may have a signifi-

cant effect on the cost-effectiveness of a health care operation. It can even be said that with

the currently poor quality standards in health care improvements in the quality of care will

typically lead to better cost effectiveness [1, 11, 12].

With reference to medical processes there are two different basic ways that the problem of cost

reduction and cost-effectiveness can be tackled.

On the one hand, the cost of the existing medical practice can be reduced by the investment in

question without changing medical processes. Examples would be cheaper drugs or cheaper an-

cillary services. This is a simple approach where the prospective ROI is easy to anticipate.

Therefore, it carries only a limited risk. But it typically yields only limited benefits and the ef-

fect may not be sustainable over longer periods of time.

On the other hand, an investment can be used to change the entire process of care or clearly de-

fined subprocesses. Examples include clinical pathways, electronic physician order entry sys-

tems, clinical decision support systems, and evidence based guidelines. These investments may

help to reduce complications, medical errors, or sort out unnecessary services [2, 3, 13]. But

they lead to complicated changes in the processes of health care delivery. They require the com-

mitment of all health care professionals involved. Therefore, the actual ROI in these cases is

difficult to anticipate, although many clinical studies have shown that they can have the by far

largest impact on cost effectiveness [14, 15, 16, 17]. In other words, investments in changing

the process of care carry a significant risk, but may lead also to a significant ROI that may be

sustainable over time.


In any case, it needs also to be assessed whether reduction in staff workload can also be trans-

lated into immediate cost savings. It may not always be possible close beds or to lay off staff

after the changes have been implemented.

The process of ROI analysis

There are different approaches to ROI analysis using different models and assumptions. The

discussion of all methodologies goes beyond the scope of this article. Rather than giving an in-

depth theoretical discussion of models and methods this chapter will focus on the practical steps

of a ROI analysis from the perspective of a health care professional. There are five steps that

appear necessary to complete a ROI analysis:

1. Analysis of the cost structure of the affected processes.

2. Assessment of the anticipated changes resulting from the investment.

3. The actual cost of the new investment.

4. The potential savings from the investment including a risk analysis.

5. Calculation of the return on investment.


Analysis of the cost structure

The analysis of the cost structure in conjunction with the ROI analysis needs to start with the

identification of those areas in the health care organization that are affected by the investment.

For instance, the use of a continuous blood gas analysis system substituting or supplementing

the benchtop gas analyser will affect the workflow of physicians and nurses on the respective

care unit. In a more complex example an electronic physician order entry system will affect phy-

sicians and nurses working with the system, the hospital pharmacy, the accounting department,

and all ancillary services that will be covered by the respective scheduling system.

This list already shows the significant task that a ROI analysis may present. All cost structures,

including variable and fixed costs, need to be identified and measured. This may be impossible

in many institutions, because the necessary cost accounting methodologies may not be in place.

Ideally, activity based costing does not only provide a detailed cost structure of the entire proc-

ess of care, but may help in itself to improve cost effectiveness [18].

What changes can be anticipated

The next step is the analysis the new investment’s effect on the workflow. This analysis may be

an integral part of the first step. One example is a clinical information system (CIS) on an in-

tensive care unit (ICU). A CIS will significantly change medical documentation, data presenta-

tion on rounds, and the nursing workflow, to name only a few areas of change. A physician or-

der entry system may even go further. It changes the way physicians write orders, how all orders

are transcribed, checked and delivered by the pharmacy, how the nurses will fulfil and docu-

ment the order.

It is important to check whether the anticipated changes will actually happen. Many of the an-

ticipated changes depend on the commitment of the health care professionals affected by the


changes. If the expectations toward the changes are unrealistic, or changes do not materialize

due to lack of commitment, the entire ROI analysis may fall apart. In this respect it should be

noted that any project involving changes of the process of care should have clinical champions,

typically respected and well accepted leaders, that lead and carry forward the implementation

of the changes.

The example of documentation with a CIS also shows that many changes may be overlooked,

as they are only indirectly associated with the process of care. Procedures that belong to the

process of documentation, for instance, are the preparation of new paper forms every day,

searching and retrieving charts, finding a place to review paper forms, performing calculations,

and communicating patient-related information among caregivers [19].

The cost of the new investment

The cost for a new investment can always be split into two groups: initial investment and run-

ning cost.

The initial investment typically includes the capital investment for the product that is purchased.

This can be the cost for hardware and/or for software. Moreover, the manpower for the imple-

mentation of the new device or system needs to be calculated. This manpower often has to be

provided by the hospital and may include efforts like the configuration of medical software or

the training of users. Depending on the financial model used in the analysis this may be a one-

time up front cost or depreciated over a number of years.

The running costs include items like consumables, energy, maintenance costs, manpower to

support and operate the systems or devices.


Table 1 gives a hypothetical, simplified example for a CIS on an ICU. It should be noted that

this example does not include reserves for future replacement of hardware. Consumables (print-

er paper, backup tapes, etc.) and energy cost are not included, too. But even this simple, and

probably incomplete, calculation shows that the list price for the CIS of EUR 300,000 represents

a total investment of close to 1M EUR over five years, and therefore, is much higher than most

health care professionals may think.

Estimation of potential savings

The identification of cost savings in the processes affected by the investment can be considered

the most difficult step in the ROI analysis. There is some level of uncertainty in assumptions

about potential changes in the processes of care before they can actually be observed. This may

be relatively easy with overt cost reductions, e.g., a new medical device uses less consumables

at a clearly defined rate. The estimation of potential cost savings is much more uncertain and

more difficult if it has to be based on assumptions about changes in the process of care that are

expected to lead to consecutive changes in outcome, e.g., less adverse drug events or less com-

plications through the use of clinical decision support.

All of the latter assumptions are prospective. These assumptions are typically based on studies

that investigated the same or similar devices or systems that are about to be assessed in the ROI

analysis in question. The problem is that these studies most often are not comparable to the spe-

cific hospital situation to be analysed [15, 20]. Some publications may even lack information

about the target variables needed in a ROI analysis.

Therefore, potential sources of variation or uncertainty should be identified and included in the

final analysis. The range of variation of the different target variables can be used to construct

best-case and worst-case scenarios. Moreover, probabilities can be assigned to the degree of


variation, so that the likelihood of the different scenarios, and therefore, the likelihood of ROI

over time can be assessed.

In general, the deeper the changes are to the process of care as a consequence of the investment,

the more difficult it will be to make a reliable estimate of the prospective ROI.

Calculate the return on investment

If the first four steps of our analysis lead to clear results the final step is relatively easy. In es-

sence, at this step we have to calculate the sum of all direct and indirect savings and subtract all

investment cost and running cost over a defined period of time. This is also the last chance to

check again for uncertainties and errors in the previous analyses, especially for hidden cost. The

final result of our calculations will be the return on investment. In any case the final analysis

should include an assessment of the probability of the results. East gives a good overview and

practical example how the quite involved methodology of probabilistic modelling can be used

in ROI analyses in health care [21].

Example: CIS for an SICU

In this hypothetical example we will expand on the previous example. The scenario is based on

the situation of a 16-bed surgical intensive care unit (SICU) in a tertiary referral centre. Occu-

pation is 90+% over the year. About 5,000 care days are done for 800 patients every year. Av-

erage TISS per day and patient is close to 40 points. 60% of the patients are ventilated.


A clinical information system is planned for this SICU. The product has already been chosen

and will include the following functionalities:

• Complete patient chart with automatic data transfer from monitor, ventilator, and IV devices.

• Complete intake & output and medication documentation with all necessary calculations.

• Medication orders with automatic reminders for due medications, but not a full order entry

system with clinical decision support.

• Plans of care, but no clinical pathway application.

• Export of all data, including QA data, scoring, diagnostic and procedure codes.

We analyse the ROI over 5 years for this CIS on our hypothetical SICU in 5 different settings:

1. Standard US hospital in a mixed reimbursement environment (capitation, case base, fee for

service).

2. Standard US hospital in the same environment but it will be assumed that the rate of medi-

cation errors is reduced.

3. Standard German hospital under the year 2000 reimbursement scheme, a mix of case base

and per diem. A ceiling for the number of patients per year is in place.

4. Same German hospital, but additional DRG and quality control data are needed.

5. Same as above, but it will be assumed that the rate of medication errors is reduced.

The analysis will follow the five steps outlined before.


Analysis of the cost structure

Without going into the exact details of a cost structure analysis there are some important differ-

ences between our hypothetical US hospital and the hypothetical German hospital:

• The physician salaries are at least twice as high in the US [4].

• Physician overtime is not always paid in German hospitals.

• Nurse-to-patient ratio in the US is higher compared to Germany.

• Malpractice premiums are 100 to 500 times higher in the US for hospital employed physi-

cians.

• In Germany a medical order is directly written by the physician into the nurse’s worksheet.

It will typically not be checked by the pharmacy. Therefore, transcription of orders by a tran-

scriptionist and checking and dispensing of orders by the pharmacy do not happen our hypo-

thetical German hospital.

• Overall hospital cost in absolute monetary terms, i.e., charges against the payer, is signifi-

cantly lower in Germany compared to the US.

How these assumption will influence the ROI analysis will be shown later on.

Anticipated changes

For the sake of simplicity we will assume that the changes to the medical workflow will be about

the same in the different scenarios. The most important changes we anticipate are:


• The nursing time spent on documentation will be reduced by about 10% as suggested by sev-

eral US studies [22, 23], although we can expect that savings are less in Germany as the doc-

umentation load is not that extensive as in the US.

• The physician time spent on documentation and retrieval of patient data will also decrease

slightly.

• Productivity gains for all health care professionals in the SICU will be about 5-10%, i.e., they

can absorb a 5-10% higher medical workload. This has been made likely in US studies [24].

While this can be translated into higher case numbers in the US hospital, the German regu-

lated health care system prevents this.

• It is assumed that data capture is much more comprehensive, so that more cost items or

charges can be captured [25].

• In two of the final five scenarios it will be assumed that the rate of medication errors on the

administration stage will by reduced by 10% [11]. The resulting savings from reduced LOS,

less medications, avoided malpractice cases, etc., will be estimated at EUR 100,000 for the

US ICU and EUR 50,000 for the German ICU.

• It is also assumed that all data can be exported to and directly be used by other information

systems in the hospital for secondary data analysis.

• No paper charts will be needed any more.

It is expected that the staff will experience a steep learning curve over the first 6 to 12 months

after implementation before the first measurable changes will actually take place [19]. The full

productivity of the CIS will be reached in the second and third year of use.


Cost of the CIS

It will be assumed that the cost for all scenarios are identical. They are based on the data in

table 1. For the calculation costs for consumables, e.g. paper, toner, backup tapes, are added at

a flat rate of EUR 10,000 per year.

Two cost scenarios will be calculated:

• Cost scenario A: All initial cost will be paid directly from the hospital budget in the year of

purchase. No credits will be taken.

• Cost scenario B: The initial cost for hardware and software will be covered by a credit paid

back over 5 years. All other initial cost will be paid from the hospital budget in the year of

purchase.

Of course, our assumption that the initial cost will be the same in all scenarios cannot hold, be-

cause of different HW and SW prices in different countries (US typically being the cheapest),

changes in exchange rates, etc. But in order to make the model not too complicated we will stick

with this assumption.

Estimation of potential savings

The potential savings greatly differ between the 5 scenarios in nearly every respect. As already

mentioned before the cost and salary structure is significantly different between the US and Ger-

many. Therefore, overtime savings will be lower in absolute financial terms in Germany. The

most important differences between the scenarios are as follows:

1. Scenario 1: US hospital


• Significant savings in overtime and reduction of additional staff, e.g., transcriptionists. These

savings will reach about EUR 200,000 in the final year.

• Further savings result from automatic capture of QA data (up to EUR 20,000 in the final

year).

• LOS reduction and productivity increase can be translated into higher patient throughput,

i.e., higher number of billable cases.

• Better charge capture will lead to higher reimbursement per case. The two latter points trans-

late into increased revenues for the hospital. Revenues will increase by EUR 400,000-

500,000 in the final year.

• Savings for malpractice premiums will account for another EUR 10,000 in the final year.

2. Scenario 2: US hospital, additional reduction of medical errors considered

• In addition to scenario 1 EUR 100,000 are saved in the final year.

3. Scenario 3: German hospital, reimbursement practice as of 2000.

• Overtime and staff savings are significantly lower than in scenarios 1 and 2 due to the differ-

ent cost structures.

• Malpractice premiums are not an issue.

• Gains in productivity and reductions in LOS cannot be translated into higher case numbers,

because the allowable number of cases is regulated. Therefore, revenues will not increase,

also because most individual charges are not reimbursed for hospitals in Germany.

• QA and DRG data (year 2000) are not legally required.


4. Scenario 4: German hospital, QA and DRG data must be collected.

• QA and especially DRG data capture is now (year 2001) legally required. Without a CIS for

each task 1/2 FTE will be needed. With the CIS close to all data can be captured without ad-

ditional workload.

5. Scenario 5: German hospital, in addition to scenario 4 the reduction of medical errors is con-

sidered.

• In addition to scenario 4 EUR 50,000 are saved in the final year.

Even before we finish our analysis, we already can see how complicated and uncertain the es-

timation of benefits and savings from an investment can be. It also shows the problem of trans-

ferring study data from one country to the other.

Calculate the ROI

The final results of the ROI analysis are displayed in figure 1 and figure 2. Figure 1 shows the

results under the assumption that all initial cost is paid directly from the hospital budget, while

in figure 2 the assumption is made that hardware and software are financed with bank credit

over 5 years.

In each of the scenarios the ROI is worse for the credit case. But this model does not account

for the fact that the budget saved by the credit may also generate return or interest which may

even outweigh the cost for the credit.

Scenarios 1 and 2 each break even irrespective of the mode of financing in the second year and

have cumulative net profits or ROI of roughly 2.5 M EUR in the final year. It should be noted

that more than half of the entire return in both scenarios stems from increased revenues.


Increased revenues are de facto excluded in the German scenarios. Also the savings in absolute

terms are only about half compared to the US scenario. Therefore, it comes to no surprise that

only in scenario 5 under the assumption of all positive effects the system barely breaks even in

the final year. In the other two scenarios the CIS is a money loser.

In other words, for the US hospital the CIS even without full order entry and clinical decision

support can be recommended from a business perspective. For the German hospital the ROI

analysis tells us that a CIS is not a good investment in financial terms. But it could well be that

a CIS that (a) includes full physician order entry and clinical decision support and (b) covers the

entire continuum of care will lead to significantly larger savings from reduction of errors [11,

26, 27]. And the situation further changes as soon as DRGs become the only reimbursement

method for German hospitals, which is expected to happen in 2003 or 2004.

Discussion and conclusions

The increasing pressure on health care expenditures and cost effectiveness of care has made

ROI analysis more important for the health care professional [2, 4, 5].

ROI analysis in health care should follow the same principles and use of the same methods as

in other industries [8]. But these methods have to be adapted to the specificities of the health

care environment. In health care we can expect more intervening factors and especially more


constraints. Therefore, existing models cannot directly be transferred to the health care environ-

ment.

For a purely technical investment, e.g., replacement of one device by the other, the ROI analysis

is relatively simple, because the process of care as such will not be changed. As soon as the in-

vestment leads to a change in the process of care, e.g., by implementing an electronic physician

order entry system, the uncertainty in the ROI analysis increases dramatically.

Generally speaking in health care the emphasis is on cost reduction rather than on increasing

revenue; an aspect that has to be included in any ROI analysis. The highest potential for cost

reduction can be anticipated when investments are used to significantly change and control

medical processes [1, 3, 16]. On the other hand, the more complex the changes to the processes

of care are, the more difficult ROI analysis becomes. It may even be difficult to assess the ROI

retrospectively.

It can be recommended to follow the five steps outlined here for the ROI analysis. Probabilistic

models can be very helpful to accommodate uncertainty in the analysis [21]. Of course, the ROI

requires that the anticipated changes are actually implemented consistently in daily practice.

The first two steps of the ROI analysis can in themselves be very useful to learn more about the

actual processes and cost factors in a given medical institution. This knowledge may even be

used to improve cost structure and cost effectiveness without additional investments. It is also

important to remember that a ROI analysis in itself requires a certain investment in manpower

and money.

Many analyses have to be based on assumptions derived from studies in the literature. It needs

to be carefully analysed whether these assumptions are actually justified in the institution under

consideration [20, 28]. Differences in the health care systems and in how medicine is practised


may make it even impossible to transfer study results from one country to the other. A rule of

thumb may be that the closer the observed changes are to the patient, i.e., the more “medical”

they are, the more likely it is that the results can be applied to different health care systems.

For example, physician order entry systems and clinical decision support systems have been

shown in U.S. studies to reduce the rate of medical errors [26, 27]. As medical errors are a prob-

lem in all developed countries and their underlying causes are comparable, it can be expected

that these study results also apply to other countries. But it should also be noted that the absolute

ROI may be distinctly different, due to different cost structures, ancillary workflows, and liti-

gation practices. On the other hand, ROI from a documentation system that captures physician

charges is unlikely to happen in other health care settings.

Our example shows how much the ROI depends on the specific health care system and setting.

And these can change over time, which makes it even more difficult to do long-term analyses.

Most of the published ROI and cost-effectiveness studies come from the US. The European

health care professional must use great caution when applying these results to her/his own anal-

ysis. Especially for information systems of any kind a much lower ROI can be expected in Eu-

ropean countries.

In summary, ROI analysis is feasible even in the highly complex setting of a health care insti-

tution. It is even necessary to justify the significant investments into to medical information

technology that will be needed in the upcoming years to cope with the problems of cost reduc-

tion, medical errors, and quality of care [29]. But even on a limited scope ROI analysis is a for-

midable task that carries a high level of uncertainty. But it is also a chance to learn more about

the processes and cost structure of care.


Acknowledgements

The author is indebted to Jay Butterbrodt, BS MBE, MBA, North Andover, Massachusetts, for

his critical review of the manuscript.

References

1. Committee on Quality of Health Care in America, Institute of Medicine. Crossing the Qual-

ity Chasm: A New Health System for the 21st Century. Washington (DC): National Acad-

emy Press; 2001.

2. Imhoff M. Clinical pathways as a tool for „intelligent“ rationing in intensive care. In: Vin-

cent JL, editor. Yearbook of intensive care and emergency medicine 2000. Springer, Hei-

delberg:Springer; 2000. p. 781-92.

3. Kohn LT, Corrigan JM, Donaldson M, editors. To Err Is Human. Building a Safer Health

System. Committee on Quality of Health Care in America. Institute of Medicine. Washing-

ton (DC): National Academy Press; 2000.

4. OECD. OECD Health Data 1999: A Comparative Analysis of 29 Countries. OECD; 1999

5. Rosenstein AH. Measuring the benefits of clinical decision support: return on investment.

Health Care Manage Rev 1999;24:32-43.


6. The National Library of Medicine. MEDLINE. Available from: URL: http://www.nc-

bi.nlm.nih.gov/PubMed (last accessed: August 11, 2001).

7. The Merriam-Webster Dictionary (1994). Springfield, MA. Merriam-Webster.

8. Karlsson G, Johannesson M. Cost-effectiveness analysis and capital costs. Soc Sci Med

1998;46:1183-91.

9. Rosenstein AH. Measuring the benefit of performance improvement and decision support.

Am J Med Qual 1999;14:262-9.

10. White J. Targets and systems of health care cost control. J Health Polit Policy Law

1999;24:653-96.

11. Bates DW, Spell N, Cullen DJ, Burdick E, Laird N, Petersen LA, et al: The costs of adverse

drug events in hospitalized patients. Adverse Drug Events Prevention Study Group. JAMA

1997;277:307-11.

12. Johnston ME, Langton KB, Haynes RB, Mathieu A. Effects of computer-based clinical de-

cision support systems on clinician performance and patient outcome. A critical appraisal

of research. Ann Intern Med 1994;120:135-42.

13. Axene D, Doyle R. Analysis of medically unnecessary inpatient service. Research Report.

Seattle (WA): Milliman & Robertson; 1994.

14. Evans RS, Pestotnik SL, Classen DC, Clemmer TP, Weaver LK, Orme JF Jr, et al. A com-

puter-assisted management program for antibiotics and other antiinfective agents. N Engl

J Med 1998;338: 232-8.

15. Gebhart F. VA Facility Slashes Drug Errors Via Bar-coding. Drug Topics 1999;1:44.


16. Leape LL, Bates DW, Cullen DJ, Cooper J, Demonaco HJ, Gallivan T, et al. Systems anal-

ysis of adverse drug events. ADE Prevention Study Group. JAMA 1995;274:35-43.

17. Teich JM, Glaser JP, Beckley RF, Aranow M, Bates DW, Kuperman GJ, et al. The Brigham

integrated computing system (BICS): advanced clinical systems in an academic hospital

environment. Int J Med Inf 1999;54:197-208.

18. Baker JJ.Activity-based costing for integrated delivery systems. J Health Care Finance

1995;22:57-61.

19. Butler MA, Bender AD. Intensive care unit bedside documentation systems. Realizing cost

savings and quality improvements. Comput Nurs 1999;17: 32-8.

20. Eckhart JG. Costing out nursing services: examining the research. Nurs Econ 1993;11:91-

8.

21. East TD. Return on investment (ROI) analysis in the face of uncertainty. Proc AMIA Symp

2000: A.

22. Goss E, Buelow V, Ramchandi H.The real impact of point-of-care computer systems on

ICU nursing costs. Journal of Systems Management 1995;46:43

23. Weiss DA, Hailstone S. Hospital saves with bedside point-of-care system. Comput Healthc

1993;14:28-32.

24. Lower MS, Nauert LB. Charting: the impact of bedside computers. Nurs Manage

1992;23:40-4.

25. Grewal R, Reed RL. Accurate charge capture and cost allocation: Cost justification for bed-

side computing. Proc AMIA Symp 1994:112-6.


26. Bates DW, Leape LL, Cullen DJ, Laird N, Petersen LA, Teich JM, et al. Effect of compu-

terized physician order entry and a team intervention on prevention of serious medication

errors. JAMA 1998;280:1311-16.

27. Bates DW, Teich JM, Lee J, Seger D, Kuperman GJ, Ma’Luf N, et al. The impact of com-

puterized physician order entry on medication error prevention. J Am Med Inform Assoc

1999; 6:313-21.

28. Gyldmark M. A review of cost studies of intensive care units: problems with the cost con-

cept. Crit Care Med 1993;23:964-72.

29. Imhoff M, Webb A, Goldschmidt A, on behalf of the ESICM. Health Informatics. Intensive

Care Med 2001;27:179-86.


er s

Initi

Hard

Soft

Tota .00

Imp 00

Use 00

Tota .00

Run

Hard .00

Soft .00

Sys .00

Tota .00

Tota .00

Tables

Table 1 Hypothetical example of the investment cost for a CIS for a 16-bed

ICU. All cost in EUR.

Title Description Value Cost ov5 year

al Cost

ware Server and bedside terminals, network including in-

stallation

100,000.00

ware License fee for 5 years including interfaces (bed-

side, HIS, LIS), cross patient analysis tools, etc.

200,000.00

l SW & HW Assumption: Credit taken for this investment, 6%

annual interest rate, 12 rates/a, linear redemption

(fixed interest rate) over 5 years

300,000.00 348,000

lementation Software implementation requires on-site configura-

tion by health care professionals. Total effort esti-

mated at 1.5 manyears (60.000 EUR per manyear).

Will be paid from hospital budget.

90.000 90.0

r Training It is assumed that 50 users are initially trained, 10

hours per user, 5 users per group. Training is done

during regular working hours.

500 hours lost working time (1 hour = 20 EUR)

100 hours working time for trainer (1 hour = 30

EUR)

13.000 13.0

l Initial Cost 451.000

ning Cost

ware Maintenance Full HW maintenance with 4 hours reaction time, as-

sumed to include spare parts.

10,000.00/a 50,000

ware Maintenance Full SW maintenance with 30 min reaction time and

remote support, assumed to include updates and

upgrades. Rate would be 15%/a of the license fee.

30,000.00/a 150,000

tem Administrator It is assumed that 1 FTE is needed for system ad-

ministration, ongoing configuration and ongoing us-

er training.

60,000.00/a 300,000

l Running Cost 500,000

l System Cost 951,000


Figure Legends

Figure 1

Return on investment over 5 years for 5 different scenarios.

All initial cost paid directly

Details in text.

Figure 2

Return on investment over 5 years for 5 different scenarios.

Initial HW and SW cost paid by credit over 5 years.

Details in text.


�

��

Figures

.

Figure 1 Return on investment over 5 years for 5 different scenarios.

All initial cost paid directly

Details in text.

Year

54321

Net

Los

s, N

et P

rofit

[M

io E

UR

]

2,5

2,0

1,5

1,0

,5

0,0

-,5

Scenario

��

��

��

��

��

��


�

��

Figure 2 Return on investment over 5 years for 5 different scenarios.

Initial HW and SW cost paid by credit over 5 years.

Details in text.

Year

5,004,003,002,001,00

Net

Los

s, N

et P

rofit

[M

io E

UR

]

2,5

2,0

1,5

1,0

,5

0,0

-,5

Scenario

��

��

��

��

��

��


How to Measure Return on Investment - Semantic Scholar...M. Imhoff: How to Measure Return on Investment 7 As it is impossible to cover all perspectives of ROI analysis in health care

Documents