Application 3 – Annotated Bibliography

InformationTechnologyPayoffintheHealth-CareIndustry-ALongitudinalStudy.pdf

Home >Information Systems homework help >Application 3 – Annotated Bibliography

Information Technology Payoff in the Health-Care Industry: A Longitudinal Study '

SARV DEVARAJ AND RAHV KOHLI

SARV DEVARAJ is Assistant Professor ofManagement at the University of Notre Dame. He received his Ph.D. from the University of Minnesota. He has worked on several projects for companies examining the impact ofthe technological environment, just- in-time manufacturing, work teams, and perfonnance evaluation systems on manu- facturing performance. He has also studied the impact of technology investments and technology usage in the health-care industry. In the field of service quality, Dr. Devaraj conducts research on consumers' perception of service and product quality in the automotive industry. He teaches courses in management of technology, operations management, and business statistics.

RAJIV KOHLI is the Project Leader, Decision Support Services, at the corporate office of Trinity Health. He is also an adjunct Assistant Professor at the University of Notre Dame and has taught at the University of Maryland, University College, where he was a recipient of a Teaching Recognition Award. Dr. Kohli received his Ph.D. from the University of Maryland. His research has been published in Decision Support Systems, Information Processing and Management, Intemational Transactions in Operational Research, Journal of Decision Systems, and Health Care Information Management, among other joumals. Dr. Kohli's research interests include organiza- tional impacts of information systems, process innovation, and enhanced decision support systems.

ABSTRACT: With the enormous investments in Information Technology (IT), the question of payoffs from IT has become increasingly important. Organizations con- tinue to question the benefits from IT investments especially in conjunction with corporate initiatives such as business process reengineering (BPR). Furthermore, the impact of technology on nonfmancial outcomes such as customer satisfaction and quality is gaining interest.

However, studies examining the IT-^erformance relationship have been far from conclusive. The difficulty in identifying impacts from technology has been the iso- lation of benefits of IT from other factors that may also contribute to organizational performance. Furthermore, benefits from technology investments may be realized over an extended period of time. Finally, IT benefits may accrue when they are done in concert with other organizational initiatives such as business process reengineering. This calls for studies that take into account control variables as well as data that span time periods. In this study, we examine monthly data collected from eight hospitals over a recent

three-year time period. We specify propositions that relate investments in IT to per- formance, and the combined effect of technology and BPR on performance. We draw

Journal of Management Informalion Systems I Spring 2000, Vol. 16. No. 4. pp. A1-67.

42 DEVARAJ AND KOHLI

upon the literature in health-care management to incorporate appropriate control variables in the analyses. Our results provide support for the IT-^erfonnance rela- tionship that is observed after certain time lags. Such a relationship may not be evident in cross-sectional or snapshot data analyses. Also, results indicate support for the impact of technology contingent on BPR practiced by hospitals.

K E Y WORDS AND PHRASES: business process reengheering, health-care infonnation systems, infonnation technology payoff, information technology productivity.

CHANGES IN THE HEALTH-CARE BUSINESS RESULTING FROM CAPITATION and de-

clining reimbursement have led to cost-cutting measures through improved opera- tions. In some cases, failure to cut costs threatens the fmancial viability of health- care organizations. While on the one hand investment in IT is seen as an enabler of efficiency and competitiveness, it is also a significant fmancial investment that, if not linked to improved organizational performance, can hasten the decline of an organi- zation. Given this scenario, the issue of IT payoff has come under close scmtiny.

While IT payoff has long been a subject of research and intense discussion, the results have been far from conclusive [1,63]. Payoffs from IT have been and continue to be open to debate in the literature, where it is called the "IT productivity paradox" [9, 12, 14, 74]. An issue that has led to the ongoing debate is inadequate methodol- ogy applied in researching the IT payoff. Perhaps the most serious issue in measuring organizational performance resulting from IT has been that IT payoff is considered in isolation and separate from other organizational practices.

To some extent, the success of IT implementation is contingent on the organization's environment, such as quality indicators resulting from process redesign initiatives, in addition to financial indicators. To this effect we will demonstrate this contingency perspective using the nature and extent of process-redesign initiatives within health- care organizations. This study aims to synthesize the literature from IT effectiveness and process redesign by developing a set of propositions. Through a rigorous analysis, we will test these propositions using empirical data collected for the purpose from a set of organizations. This study proposes to contribute to the literature by investigating pay- offs from IT investments over time, impact of IT investments on quality indicators, and impact of process redesign and IT investment on both profitability and quality.

Literature and Research Propositions

THIS SECTION PRESENTS AN OVERVIEW OF THE RELEVANT LITERATURE to frame our propositions linking technology and process reengineering with profitability and quality. We also draw on contingency theory to support the combined effect of technology and process reengineering on organizational performance. Based on the literature review and the empirical data collected, we test our data against four propositions.

IT PAYOFF IN THE HEALTH-CARE INDUSTRY 43

Technology-Profitability Connection—^The "IT Paradox"

Table 1 lists selected studies that have measured the impact of IT on profitability. The studies differed in the level at which data were collected for analysis—that is, the economy level, industry level, or firm level. At the economy level, Baily [4] found that productiv- ity declined in the 1970s, while it grew in the prior two decades. Interestingly, the decline took place during a period that incurred a significant investment in IT. A similar pattem of decline was observed during the 1970s through 1992. Roach [55] con- firmed this decline in productivity by measiuing the productivity of information workers against that of production workers. While the output of production workers increased by 16.9 percent during the 1970s through the mid-1980s, the output of information workers dropped by 6.6 percent during the same period. In terms of investment, this trend is discouraging because of the high cost of information work- ers and the steady growth in their ranks over production workers. In the manufactur- ing segment of the economy, it was determined that for every dollar invested in IT, a $0.80 margin was realized [48]. A later study at the industry level also foimd that IT investment does not lead to greater productivity than other types of investment [48].

Two studies from Asian countries show some positive results in productivity re- sulting from IT investment. Tam found that from 1983 to 1991, one of the three researched countries' economies had positive and excess retums resulting from in- vestment in IT. However, productivity was measured by increase in computer capital stock [65]. The results of a study of 12 Asian-Pacific countries found significant payoff resulting fi-om IT investment and challenge the "IT paradox." This study, conducted during 1984—90, indicates significant positive correlation between growth in IT investment and growth in both GDP and productivity. It finds that those coun- tries with higher growth rates in IT investment achieved consistently higher growth rates of GDP and productivity [40].

It should be noted that, although several studies are listed as industry-level, they were conducted across several industries, thus encompassing a significant part of the economy. The industry-level studies have also found mixed results of payofiFfrom IT investments. While some studies found that IT investment, productivity, and growth were positively correlated [35,39,42], others found no significant relationship [8,39].

The results of the firm-level studies generally show a positive IT-productivity relationship. As at the economy and industry-level studies, some firm-level studies also show no evidence of IT payoff. In two such studies Strassman foimd no correla- tion between IT investment and productivity or profitability [63, 64]. Sometimes, even when IT spending is shown to improve intermediate variables of organizational productivity such as improved communication leading to the need for reduced in- ventories [20], it does not necessarily lead to improvements in productivity [7]. However, in a firm-level study Brynjolfsson found that firms that reengineered were significantly more productive than their competitors [10].

In a study of U.S. retail banking, IT capital investment was found to have no real benefits. However, this study concluded that there were high retums from investment in IT labor [52]. A recent study has found that investments in IT capital are a net

44 DEVARAJ AND KOHLI

Table 1. A Summary of Selected Studies at the Economy, Industry, and Firm Level, and Key Findings

Study

Baiiy [4] Roach [55]

Morrison and Bemdt [48] Kraemer and Dedrick [40]

Jorgenson and Stiroh [33]

Tam [65]

Siegel and Griliches [60]

Brendt and Morrison [8]

Kelley [35]

Lehr and Uchtenberg [42] Koski [39]

Strassmann [64]

Dudley and Lasserre [20]

Strassmann [63]

Banja, Kriebel and Mukhopadhyay [7]

Diowert and Smith [19] Hitt and Brynjolfsson [30] Brynjolfsson [9]

Prasad and Harker [52]

Dewan and Min [17]

Mukhopadhyay, Rajiv and Srinivasan [49]

Level of study

Economy Economy

Economy

Industry

Industry/fimi

Firm

Firm Firm

Firm

Firm/appiication

Result

4. t

i ^

4.4.T

t 1̂

<-»

<->

«->

«-»

t t

Key fmdings

Productivity growth declined in the 1970s Output of the information work dropped while that of the production work increased during the

odiiit? pmiuu IT does not lead to improved profitabiiity IT investnnent in 12 Asia Pacific countries had a significant positive correlation with growth in both GDP and productivity between 1984 and 1990. Productivity growth dropped from 1.7% per year for the 1947-73 period to about 0.5% for the 1Q71_jQ9 rmrinri I9/O—9£ pBriou IT investment in firms led to positive and excess returns (1983-91) in one of the three economies (Hong Kong, Singapore, and Malaysia) IT investment and productivity growth have a positive correlation IT investment does not lead to any greater productivity then other types of investment IT investment (in programmable automation technology) leads to significant efficiency

dU Vdi 1 lay o IT intensity and productivity growth during the period 1987-82 are strongly correlated IT investment by firm's use of advanced communications technologies and its output and productivity have no positive relationship No correlation between IT and return on investment IT spending to improve communication and information reduce the need for inventories. Profitability link was not mentioned. No relation between IT spending, profits, and productivity IT was positively related to some intermediate measures of profitability, but that the effect was generally too small to measurably affect final oiitniit uuipui. IT led to large productivity gains IT led to increased productivity and consumer surplus, but not higher profitability Firms that had reengineered were significantly more productive than their competitors Additional capital investment in IT may not have real benefits IT capital is a net substitute for ordinary capital and labor, i.e., IT investment leads to higher returns IT investment leads to higher productivity and quaiity

substitute for ordinary capital and labor capital, that is, IT investments lead to greater returns as compared to same level of investment in other forms of capital [17]. An- other recent study at the application level of a firm in the U.S. Postal Service found that investments and use of IT led to higher output of mail-sorting facilities [49]. This

IT PAYOFF IN THE HEALTH-CARE INDUSTRY 45

points to a key nuance that investments in IT have to be coupled with the actual use of IT. It is usually assumed, perhaps erroneously, that increased investments in IT lead to increased usage of IT.

The literature on the subject of the IT paradox and why IT has not shown signifi- cant value is comprehensive and provides a basis for explaining some of the incon- sistency and also offers valuable guidance for future research. Research designs have varied from a cross-section of performance variables to longitudinal data. The dura- tion of data collected and the design of studies also vary significantly. Some research designs gathered data at a point in time for research relationships between productiv- ity and IT investment [53] or three to five annual data points [7, 17,30, 52]. Without sufficient control variables, annual aggregation of data may fail to account for other influences or factors that can affect the firm. Furthermore, three or five data points may not be sufficient to establish a trend for IT payoff, especially when there are lag effects resulting from IT investment and noticeable payoffs. A data set that includes quarter-level data for three to five years, combined with the corresponding control variables, would flimish sufficient data points and provide meaningful results [49].

Although the variety in variables for IT payoff adds diversity to the question of IT payoff, it is also likely to lead to inconsistent research designs, thus hampering the development of a research tradition in this growing field. Table 2 provides a sam- pling of recent firm-level studies, variables studied, the duration of the study, and the key findings.

Based on the literature presented above, the connection between IT investments and payoff is not very conclusive. This may be due to factors related to level of study, variables selected, and research design. We attempt to address some of these con- cems, beginning with our first proposition:

Proposition 1: Investment in information technology leads to increased profit- ability of the organization.

Technology-Quality Relationship

One potential explanation for the mixed results of IT payoff is that the variables collected have varied among the published studies. Financial variables such as re- tum on assets (ROA) [30] and retum on investment (ROI) have generally been the mainstay for dependent variables, while capital, labor, operating expenses, and rev- enues have been widely used as independent variables for investigating IT payoff within organizations [17, 30, 52]. For the control or independent variables, some studies have gathered operational measures such as inventory tumover [7] and asso- ciated costs [19] to study the impact of IT.

Other studies have gathered organizational variables such as absenteeism, degree of supervision [49], employee composition [24], and number of years the CIO has been in the position with the organization [53] to study the IT payoff for the firm. Even the issue of whether uniformity of variables is preferred has been debated in the IT effectiveness literature [58]. Health-care profitability studies have gathered inde- pendent variables that represent ownership type before and after the prospective

46 DEVARAI AND KOHLI V Table 2. A Summary of Firm-level Studies, Variables Used, Duration of the Study, and Key Findings

Study

Barua, Kriebel, and Mukhopadhyay \7]

Diewert and Smith [19]

Hitt and Brynjolfsson [30]

Prasad and Harker [52]

Dewan and Min [17]

Mukhopadhyay, Rajiv and Srinivasan [49]

Prattipati and Mensah [53]

Francalanci and Galal [24]

Variables used

Capacity utilization. inventory tumover. quality, relative price. and new product introduction Inventory holding costs. growth rate, purchases, sales, inventory levels Value added, IT stock. noncomputer capital. ROA, labor expense. ROE, shareholder return, IT stock/employee, capital investment, sales growth, market share. debt, R&D stock firm IT capital, non-IT capital. IS labor expense, non-IS

lauor expense IT capital, non-IT capital. labor expense, value added, sales, number of employees Total output, on-time output, labor hours. machine hours, level of automation. absenteeism rate.

uegruo 01 supBrvision Number of years CIO in the position, proportion of software resources spent on client server applications, percentage of software budget spent on new development. IT investments, clerical. managerial, and professional composition, income per employee, total operating expense

* Variable names changed to generic.

Duration

Annual over 3 years

Ouarterly over six quarters Annual over 5 years

Annual over 3 years

Annual over 5 years

39 accounting periods over 3 years

One year

10-year period

Key findings

IT was positively related to some intermediate measures of profitability, but that the effect was generally too small to measurably affect final output. IT leads to large productivity gains

IT leads to increased productivity and consumer surplus, but not higher profitability

Additional capital investment in IT may not have real benefits

IT capital is a net substitute for ordinaiy capital and labor, i.e., IT investment leads to higher retums

IT investment leads to higher productivity and quality

Highly productive firms spent more on client-server and less on in- house application development

Increases in IT expenses are associated with productivity benefits when accompanied by changes in worker composition)

payment system (PPS) [25], and not-for-profit status [16]. Past studies in health care have used a niunber of hospital-related variables in addition to fmancial measures to determine those factors that affect the profitability of hospitals. Such measures in- clude patient mix, average length of stay (ALOS) of patients, and Herfindahl index of market concentration [71]. Table 3 summarizes studies that have used organizational variables to measure IT payoffs.

The topic of quality has gained renewed importance as a management concept and is often supported by significant investments in IT. There is evidence that firms that have won quality awards outperform similar firms in operating-income-based measures. Furthermore, high-quality firms also are better at controlling costs [29]. Therefore, quality variables are often considered in research on IT payoff [50, 73].

IT PAYOFF IN THE HEALTH-CARE INDUSTRY 47

Table 3. A Summary of Studies Utilizing Oiganizational Variables to Meastire IT Payoff, and Key Findings

Study Orgatiizational

variables Key findings

Willcocks and Lester [73]

Anderson, Fomell and Rust [2]

Antonelli [3]

HendricksandSinghal[29]

Myers, Kappelman, and Prybutok 150)

Francalanci and Galai [24]

Measurement and evaluation of quality Improvement

Customer satisfaction and productivity

Diffusion of IT, effects on tfie tradability, divisibility and transportability of infonnation, and interaction between receptivity eind connectivity of leaming agents

Operating income, assets, sales, costs, quality award

Quality—service, system, information; Impact—individual, work group, organizational; Usage, user satisfaction

Worker compensation, IT expenditure

Grover, Teng, Segars, and Fiedler [28] ^

Pinsonneault and Rivard [51]

Teo and Wong [67]

fvlature and magnitude of relationships between IT diffusion and process redesign

IT investment and managerial work

Information quality and work environment

Elements in the uncertainty about the IT payoff relate to deficiencies in measurement at the macroeco- nomic level and weaknesses In organizational evaluation practice

Changes in customer satisfaction and cnanges in productivity is positive for goods, but negative for services (1) strong conflation between the levels and rates of growth in the use of communication and business services, (2)The productivity enhetncing effects of the co-evolution in the use of business and communication.

Firms that have won quality awards outperform other similar firms in operating-income based measures. Quality award winning firms also control costs better.

A comprehensive contingency model for eissessing the IS function should include environ- mental varieibles such as industry, competition, economy; and organizational variables such as IS budget, size of IS function, maturity of IS function, top management support.

Increases In IT expenses are associated with productivity benefits when accompanied by changes in worker composition Process redesign and IT have a complex relationship with productivity, and that these can be represented by a mediating or moderating nwdel for different technologies

Heavy IT users paid greater attention to and spent more time on the infonnation related Improvement in work environment is positively related to IT organizational impact but not to managerial satisfaction.

Business process redesign or reengineering (BPR), related to the quality principles, has also been examined in relation to IT spending and productivity of organizations [28].

With increased competition and the focus on satisfying customer needs, customer satisfaction is also being researched in IT payoff studies. Health-care and other ser- vice organizations do not have the luxury of traditional quality control to ensure product quality prior to delivery, as in manufacturing organizations [21]. Therefore, patients' satisfaction is surveyed to assess the quality of services. Patient satisfaction data have become more important and are now considered one of the outcomes of care. Results of patient satisfaction are also used to identify protocols of care that

48 DEVARAJ AND KOHLI

result in prefened clinical outcomes, lower costs, and the highest level of patient satisfaction [26,38].

Thus, there is reason to believe that, in addition to the impact of technology on profitability ofthe organization, the impact of technology on the quality of services rendered is of equal importance. Thus, our second proposition outlines the technol- ogy-quality connection:

Proposition 2: Investment in information technology leads to improved quality of products or services as assessed by measures of customer satisfaction and service or product quality.

Process Reengineering

While IT payoff continues to be investigated, there have been attempts to create strategies and frameworks that reinterpret or explain contradictory results from past research. Robey [56] presents a number of strategies that suggest scrutinizing the data from past studies and correlating similar studies to examine commonalties among research sites and samples. Another suggestion is to "widen the lens" and examine the results of IT payoff in the context of political theory, organizational culture, institutional theory, and organizational leaming theory. Each theory exam- ines organizational change as a process. A process theory view of IT and business value is also proposed by Soh and Markus [62], who suggest that IT use and know- how are intermediate outcomes and require further research. The process theory model suggests that investment in IT projects, applications, and skill base represent creation of IT assets in an organization. Successful deployment of IT assets leads to IT impacts such as redesigned processes, improved decision making, and improved coordination [22, 37]. It is only when IT impacts are at strategic places in the orga- nizational structure that we will see enhanced organizational effectiveness. It is this organizational effectiveness that most studies have examined through the selection of fmancial and productivity variables. Soh and Markus argue that, between the IT assets and organizational effectiveness, there can be many "losses" that prevent the organizations from realizing a payoff [62]. The process view of IT payoff is also echoed by Mooney, Gurbaxani, and Kraemer in their fi-amework proposing that firms derive business value from intermediate operational and management pro- cesses. They classify these processes along automational, informational, and trans- formational dimensions. As IT continues to permeate the organization, it has a greater impact on the processes and eventually on the organization [47].

Although most BPR is done with the customer in mind, organizations have real- ized benefits from BPR ranging from financial benefits to customer satisfaction and growth sustenance. CIGNA Corporation, a multinational, leading provider of insur- ance and fmancial services, successfully completed a number of BPR projects and realized savings of $100 million. For every $1 invested in BPR, it got back $2 to $3 in benefits. CIGNA accomplished these savings while achieving improvements in cus- tomer satisfaction and quality of services. In investigating how CIGNA achieved these results, the case study found that the success in BPR was accomplished by affecting

IT PAYOFF IN THE HEALTH-CARE INDUSTRY 49

changes in the organizational structure and its business practices. The organization redesigned and created self-focused teams and trained generalists to handle customer service. Business practices such as enterprise-wide metrics, application of skills where added value was achieved, and a common view of the customer were implemented [15].

Charles Schwab and Co., an investment broker, reengineered the "cashiering" pro- cess to standardize the flow of incoming and outgoing funds across multiple invest- ment lines [44]. Similarly, Florida Power and Light (FPL) was awarded the Deming prize for its quality initiatives and reaped significant savings by developing value- added processes [57].

Our third proposition, thus, is:

Proposition 3: Organizational factors, such as business process redesign (BPR), have a positive impact on measures of organizational profitability and quality.

Fit Between Technology and Process Reengineering

The imderlying premise in "contingency theory" is the proposition that organizational performance is the result of a "match" or 'Tit" between factors [69, 70]. Better perfor- mance is realized when there is a good fit, or congruency, between these factors, and not otherwise. In the context of technology investments, contingency theory would suggest that, while technology and organizational practices (such as BPR) may have separate impacts on performance, the two together may also affect performance significantly. In other words, the impact of technology on performance is contingent on whether other organizational processes, such as BPR, were also implemented.

Van de Ven and Drazin stated that one approach to capture the fit between factors in contingency theory is through "interaction" terms [69]. According to Venkatraman, such an interpretation of fit would be consistent with "fit as moderation" [70]. From this perspective, the impact of a predictor variable on performance depends on the level of a third variable. In our context, we posit that the impact of technology on performance depends also on the level or degree of BPR implementation in the organization.

Contingency theory and the notion of "fit" received substantial analytical rigor with Milgrom and Roberts's detailed and analytical examination of fit using the mathematics of complementarity [45]. In terms of complementarity theory, activities are complements if doing (more of) any one of them increases the retums to doing (more of) the others. Specifically, we contend that doing more of BPR increases the retums of investing in technology.

Also drawing upon notions of complementarity, Barua et al. presented a theory called business value complementarity. One of the arguments based on this theory was that investments in IT and reengineering cannot succeed in isolation [5]. Since technology and business processes were viewed as complementary factors, they must be changed in a coordinated marmer to improve performance.

The operationalization of the fit between technology and process redesign is an interesting question. The literature offers various forms or perspectives of fit and the analytical operationalization varies according to the form of fit proposed. The per-

50 DEVARAJ AND KOHLI

spective that best fits our characterization of the concert between technology and BPR is that of "fit as interaction," since we hypothesize that environments that have significant technology investments if coupled with significant BPR initiatives will restilt in improved organizational perfonnance. According to this approach, the fit between technology and BPR is tested using the cross-product or interaction be- tween the variables for technology and BPR. If the interaction term is statistically significant, after the main effect of technology and BPR is taken into account, there is evidence that the fit between these two factors matters.

Based on contingency and complementarity theory prescriptions, otu- fotuth and final proposition is:

Proposition 4: The interaction between technology investment and BPR initia- tives undertaken by the organizations will have a positive impact on perfor- mance. In other words, the impact of technology on performance is higher in organizations where there is a high degree of BPR implementation and lower in organizations where there is lesser BPR implementation.

Research Design

THE HOSPITALS INCLUDED IN THIS STUDY ARE MEMBERS OF A HEALTH SYSTEM. The

health system is a national organization with member hospitals in various markets across the United States. Each hospital is an independent legal entity, each with its own board and financial statements. The member organizations ofthe health system have combined beds of over 4,000, employ about 20,000 people, and have a total operating revenue of approximately $1.5 billion. Most of the hospitals have been providing health care for over a hundred years. The hospitals provide a range of services including acute-care hospitals, extended-care facilities, residential facilities for the dis- abled and elderly, occupational medicine, and community service organizations.

Our research design, called panel data, consists of data across hospitals as well as over time. As a meastire of investment in IT, we collected data for 36 monthly periods fVom eight hospitals of the health system that had recently implemented a decision support system (DSS) to help evaluate contracts. These contracts estimate costs of expected services and compare them with expected payments. The DSS also helps identify areas of cost-cutting and operational improvements necessary for the hospital's fmancial viability.

The cross-sectional set of hospitals combined with time-series data is ideal for examining the effect ofthe usage of IT on measures of profitability and quality, while at the same time controlling for various other factors.

Decision Support Systems (DSS)

Decision support systems (DSS) are computer systems designed to help improve the effectiveness and productivity of managers [34]. DSS deliver models that can be used systematically to evaluate policies and alternatives [22]. Conventional hospital in-

IT PAYOFF IN THE HEALTH-CARE INDUSTRY 51

formation systems help meet the challenge by providing data necessary for policy fonnation and outcome measurement. However, integrating those systems with DSS can help managers gain insight into the operations, consider alternatives, and de- velop business strategies.

Combined with cost information, DSS can be used for operational, managerial, and strategic decision-making. Operational decision-making includes resource alloca- tion, activity-based costing, and such decisions that improve the operations at the patient-care level. Managerial decision making includes cost containment, overall profitability of the department, and integration of departmental services with those of other departments. Strategic uses of a DSS can involve contracting decisions, pricing decisions, and merger/ acquisitions planning [38].

In the organizations included in our study, the DSS plays a critical role in the analysis of data and the identification of areas for BPR. The DSS serves as a reposi- tory for the financial, clinical, and quality outcomes for each patient admitted to any of the hospitals for the past several years. Each patient record contains as many as 400 fields describing the status of the patient from the time of admission to discharge. Furthermore, the DSS database is designed to store day-of-stay data; for example, the DSS database can supply infonnation that a certain number of aspirin tablets were given to a cardiac patient on the third day of stay. The DSS data are analyzed to identify "best practices" to be emulated by other hospitals. Benchmarking data are purchased from commercial organization and stored in the DSS, so decision makers can measure their performance with national and regional health care organizations.

With this level of information, the managers are able to conduct analysis of the impact of fmancial and clinical changes to the patient population. The DSS is also a competitive tool in conducting "what-if' analysis to assess the profitability of con- tracts for a given patient population.

Dependent Variables

Measures of Profitability and Quality

Following previous studies in hospital profitability [27, 41, 59, 68], we chose rev- enues as a measure of profitability. Traditionally, profitability measures call for sub- tracting costs from revenue. However, in health care this task is made difficult for the following reasons: (1) most hospitals lack accurate cost accounting systems and often use a constant ratio of cost to charges (RCC) to determine the cost of services. Thus, RCC does not facilitate the analysis of IT payoff as it is a direct transformation of revenue. (2) Costs are also affected by varying contractual agreements—that is, discounts offered by hospitals to insurers, and write-offs toward charity care. Each of these deductions can convolute the traditional profitability measures. (3) With the increase in managed health care, premium revenues are considered as profits and expenses are treated as a charge against such profits. On the other hand, revenues are not aifected by contractual agreement with insurers, patients' ability to pay, or with managed health care.

52 DEVARAJ AND KOHLI

In order to avoid any bias resulting from patients' varying length of stay, we use two measures of revenue—net patient revenue per day and net patient revenue per admis- sion. All revenue numbers were disguised by multiplying with a constant to protect the confidentiality ofthe data. The following is a brief description ofthe measures for profitability and quality:

• Net patient revenue per day (NPRDAY) is the ratio ofthe total revenue realized by the hospital to the total number of days in the period under consideration.

• Net patient revenue per admission (NPRADM) is the ratio ofthe total revenue realized by the hospital to the total number of patient admissions during the period tinder consideration.

• Mortality rates (MORT) is defmed as the number of mortalities within 30 days of an operative procedure divided by the total number of operative procedures conducted in the time period under consideration.

• Customer satisfaction (SATIS) is defmed as the percentage of "top-box" scores (described below) in the time period under consideration.

Customer satisfaction is measured by independent survey research companies through phone calls made to patients. This ensures the confidentiality of patient responses as well as maintaining the reliability ofthe data. The questions measuring satisfaction include willingness to recommend the hospital to others, satisfaction with the doc- tors and nurses, and treatment with dignity, respect, and compassion. Finally, a ques- tion solicits overall satisfaction. The correlations of all satisfaction questions with the overall satisfaction question are regularly monitored to ensure that the questions meaningfully capture patient satisfaction. The responses are ranked on a ten-point Likert scale. The patient satisfaction scores in this paper represent top-box scores—that is, the percentage of respondents ranking 9 and 10 on the ten-point scale. Top-box ranking is a widely used method for assessing patient satisfaction in the health care field.

Independent Variables

Technology Investment

We collected monthly costs associated with IT labor, capital, and support for every hospital. Specifically, we focused on the expenditure involved within the DSS.

• ITLABOR: Our measure for IT labor includes costs associated with total salary and wage expenses for management, supervisors, professionals, administrative and clerical staff.

• ITSUPPORT: The measure for IT support includes (a) consulting fees expense, (b) decision support system computer programming, (c) software support, and (d) decision support system maintenance expense.

• ITCAPITAL: IT capital expenses include cost ofthe DSS software product and its associated modules.

IT PAYOFF IN THE HEALTH-CARE INDUSTRY 53

Nature of BPR Initiatives

rII III ii Category of BPR

Figure 1. The Nature of BPR Initiatives

Implementation of Business Process Redesign (BPR)

In health care, the application of BPR has generally focused on cost reduction of clinical and administrative processes [66]. BPR can assist in reengineering patient care by identifying practice pattems that lead to reduced costs, improved quality outcomes, and higher patient satisfaction. These practice pattems, also called treat- ment pathways, in tum lead to reengineered patient care.

The BPR projects in the organization were focused in a number of areas. These can be categorized in five major areas indicated in the histogram in figure 1. Most initia- tives were for improving the quality of outcomes of hospital care. With increased cost cutting, the quality of outcomes such as the length of stay, infection rates and ex- pected versus actual mortality rates are being carefully examined. The second largest number of BPR projects were applied to improve decision-making. An example of improved decision-making is developing a treatment regimen for heart attack pa- tients that leads to the highest survival and rehabilitation rate. Cost containment initiatives include administrative as well as clinical scenarios. The best-quality and lowest-cost physicians for the expensive procedures, such as heart bypass surgery, are identified. The treatment protocols of such physicians are compared with those of other physicians. The breakdown of costs of drugs, prosthetics, and lab tests is then shared with the physician committees. Similarly, patient wait times in the emergency room were studied, and a BPR initiative to install an information system to commu- nicate lab results efficiently to the physician was undertaken.

Process improvement projects were applied in various administrative areas. Given that the reimbursement for patient services is dependent on the accurate coding of diseases and procedures, the process of coding was examined and improved for fair reimbursement from payers. A few BPR projects were initiated to explore and support new services being provided to physicians and patients, some of which had been outsourced in the past.

Over the period of study for the hospitals in our sample the maximum number of

54 DEVARAJ AND KOHLI

BPR initiatives under way was 18, the minimum was 4, and the average was 12.8. Given that our study encompassed 77 BPR projects, space considerations prevent us from providing greater detail. Additional detail on the BPR projects is available from the authors.

The following is a detailed example for establishing standards and expected cost savings for knee and hip procedures classified as a Diagnostically Related Group (DRG) 209. From the historical data, we compare actual verstis expected costs (from benchmarking) for severity-adjusted DRG 209. Then, a breakdown ofthe total costs at the departmental level with a focus on high-cost departments is presented to the process engineer. This information is fiirther broken down into individual charge item groups such as supplies, room charges, and other high-cost items. In addition, median cost by departments serving DRG 209 are also calculated. Costs above the median are identified by department and by physician(s). Finally, savings opportuni- ties are identified by bringing the practice pattems to median and benchmark levels, and patient care is reengineered. Other clinical applications of BPR have been in areas of reducing the length of stay [22] and the turnaround time of test orders be- tween the pathology laboratory and the Emergency Department [36]. Similarly, BPR has led to improvements in the billing process by reducing the number of days in Accounts Receivable and thereby improving the cash flow [23].

The empirical literature in BPR has employed various sets of measures, many of which are subjective assessments of managers obtained through surveys. Neverthe- less, a common theme among BPR studies is cost reduction and process-time reduction. Because ofthe diversity of physician practice pattems, process steps or process times in health-care organizations are studied in the light of potential cost reductions.

We conducted in-depth interviews with managers to get their assessment of an appropriate measure of BPR effectiveness. The managers' assessment indicated that, while there were various areas where improvement could be measured, the most important measure of any BPR initiative would be its "impact on the bottom line"— in other words, the anticipated cost savings to the hospital. Therefore, we obtained an estimate ofthe potential cost savings to the organization as a result of implementing a particular BPR initiative. The BPR variable used in this study for a particular hospital for a given time period is the anticipated total cost savings of all the BPR projects under way.

We test the proposition that the greater the extent of BPR, the greater the possibil- ity for improved performance in concert with technology usage (propositions 3 and 4). This measure has the advantage that it is objective and does not rely on the subjective judgment of any individual.

Lag Effects

In a longitudinal study the effects ofthe independent variables on performance may be realized after a period of time. Brynjolfsson and Hitt (1998) suggest that "if there is some lag or adjustment time required to match organizational factors and IT invest-

IT PAYOFF IN THE HEALTH-CARE INDUSTRY 55

Capital investment in hardware & software

Ubor& Support

Reports programs

Data analysis

Changes in patient

care

Financial & quality

outcomes

Figure 2. Timeline of IT Investment Impact

ments, we would expect to see more benefits over longer time periods" [11]. Simi- larly, Mahmood and Maim (1997) state, "Attempts should be made to use a time- lagged regression analysis to allow for the fact that benefits derived in a given year may be due to IT investments made in previous years" [44].

Specific lags vary based on the nature of the industry and the processes being considered. While in manufacturing and engineering applications lag effects can be measured in years, the health care industry is likely to experience shorter lags due to the nature of its business. Furthermore, the maturity of the IT infrastmcture within the organizations can also affect the duration of lags.

Our discussions with managers at the research site suggested that investments in IT follow a particular chronological sequence in the hospital setting. This is shown in figure 2. The first step is typically an investment in IT capital. This is followed by procurement of software and hardware. The new infrastmcture is then used to create and mn programs and reports. This stage involves IT labor and IT support as inputs. This is followed by member organizations (hospitals) making changes based on the results of the earlier step. This leads to changes in patient care that eventually lead to better performance on financial and quality outcomes.

Managers were also of the opinion that "investments in IT labor may yield results in terms of improved performance about 2—3 months later due to lead times involved with the technology being in place prior to the programming staff developing the applications." Thus, we chose to incorporate a time lag of three time periods for IT labor and support in our analyses. Since capital enters the equation earlier along the time line (figure 2), we included a four-period lag for IT capital.

Our measure for BPR initiatives captures the anticipated cost savings resulting from BPR implementation. The infrastmcture to support such an initiative is already in place, and months of careful plarming have already taken place. This is in contrast to startup organizations where the infrastmcture deployment can take several months or even years. Thus, in our case it seems reasonable to expect that retums from BPR may be realized without much delay. This was also supported by managers who were responsible for implementing these initiatives. One manager stated that "the im- pact resulting from the change in practice pattems will appear shortly, approxi- mately 2 months in most cases, because revenue is booked right after the patient is treated. Therefore the lags are primarily due to the physicians incorporating the new process into practice and the time involved in compiling medical and billing records." Therefore, we employed a two-pwiod lag on the BPR variable.

56 DEVARAJ AND KOHLI

BPR Studies in clinical settings suggest lag effects of four to six months. Shorter lead times in clinical settings can be ascribed to the implementation of "evidence- based practice" that promotes making immediate process changes based on pub- lished clinical evidence [61]. For example, process changes in the emergency depart- ment resulted in a reduction of laboratory utilization with a lag of six months [18]. In an intensive care tmit (ICU), respiratory itifections were reduced significantly after a lag of four months of a BPR initiative [31]. Implementation of practice guidelines to reduce lab tests. X-rays, and EKG exams resulted in improved process and outcomes in less than six months [72].

Control Variables

Conceivably, the perfonnance of hospitals can be affected by a number of variables other than the investment in IT and process redesign. Therefore, we conducted an extensive literature siurey of projects on determinants of health care productivity and profitability. The list and labeling of control variables that we employ in our study reflect the extant literamre in health care management [27, 41, 59, 68].

• Service index (CASEMIX): The service index or the casemix index is a meastu-e of the range of services offered by the hospital. The higher this measure, the more complex the services rendered by the hospital.

• Labor intensity (FTE): The number of full-time employees per patient day pro- vides a measure of labor intensity. The relationship between personnel effi- ciency and profit margins has been of interest to health-care-management re- searchers and thus should be controlled for.

The next two variables assess the extent to which services offered by the hospital were provided to Medicare and Medicaid patients. The reasons for controlling for these effects were: (1) Medicare and Medicaid payments are typically less than pay- ments from other payers for similar services, and (2) Medicaid and Medicare patients are more costly to treat than other patients [27].

• Medicare (MEDICARE): Percentage of admissions that are Medicare patients compared to total admissions;

• Medicaid (MEDICAID): Percentage of admissions that are Medicaid patients compared to total admissions;

• Outpatient mix (OUTPATNT): There is a general belief in health care manage- ment that outpatient services are more profitable than inpatient services. To control for the existence of any such effect on profitability, we include the ratio of outpatient revenue to total revenue as a control variable.'

• Per-capita income (INCOME): It is conceivable that with higher patient in- comes, hospitals may be able to charge more for their services as well as lose less due to bad debts. For this reason, we included the per-capita income of region/ market for each ofthe hospitals in the analyses.

A summary of all the variables employed in this study is shown in Table 4.

IT PAYOFF IN THE HEALTH-CARE INDUSTRY 57

Table 4. Definition of Variables

Variable name Definition

NPRDAY, Net patient revenue per day for hospital /during time period t NPRADM^ Net patient revenue per patient admission for hospital /during

period t MORT,, Patient mortality for hospital /during period t SATIS, Customer satisfaction performance for hospital /during period t ITLABOR, IT labor expenses for hospital /during period f ITSUPPORT, IT support expenses for hospital /during period / ITCAPj IT capital expenses for hospital /during period t BPR, Measure of fhe extent of anticipated BPR impact for hospital /

during time period t CASEMIX, Case mix for hospital /during time period t FTE,, Number of f ull-f ime employees at hospital /during time period t MEDICARE, Percent of Medicare admissions at hospital /during time period t MEDICAID, Percenf of Medicaid admissions at hospital /during time period t INCOME, Per-capita income of the region/market for hospital /during time

period/ OUTPATNT, Ratio of outpatient revenue to total revenue for hospital / during

time period f

Estimation

W E EMPLOY TIME-SERIES MODELS TO ESTIMATE THE EFFECT of technology and BPR implementation and their combined effect on various measures of performance. These time-series models account for longitudinal correlation or correlations over time. Equations (1) through (4) were estimated to examine the relationship between technology and performance while at the same time controlling for various extrane- ous factors through the use of control variables:

(1) NPRADM = po + p 1 ITLAB0R(3) + P2 ITSUPP0RT(3) + p3 ITCAPITAL(4)

+ p4 BPR(2) + p5 ITLAB0R(3) * BPR(2) + p6 ITSUPP0RT(3) * BPR(2) + P7 ITCAPITAL(4) * BPR(2) + p8 CASEMX + p9 FTE + p 10 MEDICARE

+ P11 MEDIC AID + p 12 OUTPATNT + P13 INCOME.

(2) NPRDAY=po + p 1 ITLAB0R(3) + p2 ITSUPP0RT(3) + p3 ITCAPITAL(4)

+ P4 BPR(2) + P5 n'LAB0R(3) * BPR(2) + P6 ITSUPP0RT(3) * BPR(2) + P7 ITCAPITAL(4) • BPR(2) +p8 CASEMDC + p9 FTE + p 10 MEDICARE

+ pi 1 MEDICAID + p i 2 OUTPATNT + p l 3 INCOME.

(3) MORT = po + p 1 ITLAB0R(3) + P2 ITSUPP0RT(3) + P3 ITCAPITAL(4) + P4 BPR(2) + P5 ITLAB0R(3) * BPR(2) + p6 ITSUPP0RT(3) * BPR(2)

58 DEVARAJ AND KOHLI

+ p7 ITCAPITAL(4) * BPR(2) + p8 CASEMIX + p9 FTE + plO MEDICARE -H pi 1 MEDICAID + p 12 OUTPATNT -H p i 3 INCOME.

(4) SATISF = pO + pi rrLAB0R(3) + p2 ITSUPP0RT(3) + P3 ITCAPITAL(4) + P4 BPR(2) + P5 rrLAB0R(3) * BPR(2) + P6 ITSUPP0RT(3) • BPR(2)

+ p7 ITCAPITAL(4) * BPR(2) + p8 CASEMIX + P9 FTE + p 10 MEDICARE + pi 1 MEDICAID + p 12 OUTPATNT + p l 3 INCOME.

Net patient revenue per day (NPRDAY) and net patient revenue per admission (NPRADM) are widely used measures of hospital profitability, while in-patient mor- tality (MORT) and customer satisfaction (SATISF) are measures of quality perfor- mance. The direct impact of investments in labor, support, and capital can be esti- mated by examining the coefficients (P1H33) associated with these terms. The impact of BPR on perfonnance is indicated by coefficient P4. Finally, the fit between IT investments in labor, support, and capital and BPR is indicated by the sign and significance of the coefficients associated with the interaction terms in the model (P5-p7). The numbers in parentheses indicate the number of time lags considered for the respective independent variables.

D i a g n o s t i c C h e c k s

We performed several diagnostic checks to ensure that assumptions of the analyses were not violated. First, we captured the residuals from the analyses and tested whether they followed a normal distribution using the Kolmogorov-Smimov test. The/) val- ues that we obtained did not indicate a violation of the nomiality assumption. The second check we performed was to test for nonconstant variance or heteroscedasticity using White's test. The results did not suggest any issues. We also ensured that autocorrelation or serial correlation was not a problem in the estimation by looking at the Durbin-Watson statistic. Our final check was to test for the presence of multicollinearity. None of the variance inflation factors (VIF) was greater than the threshold value of 10, suggesting that multicollinearity was not an issue.

Results

We present descriptive statistics and pairwise correlations between variables em- ployed in this study in Table 5. We observe several sigtiificant correlations between variables. We limit our discussion of these correlations only to key relationships because correlations hint at relationships between variables without accounting for the impact of the other variables. The measures for profitability are positively and significantly correlated with each other and also correlated with customer satisfac- tion. Investments in labor are positively correlated with support and capital, whereas support and capital are not significantly related with each other.

The relationship between IT investments (in labor, support, and capital) and profit- ability is the subject of proposition 1. An examination of model 1 (presented in Table 6) suggests that labor investment at any time period has a significant effect on net

S 0

o**—j

jr re

D A

Bi CL

c CL

o CO •* 9

cj in t̂ o o •* o h~

o d o d

n in CO m in <o IO •*

h- ^ 05 »- »- in T^ O T- C3 CM ^

d d d d d d T- 05 CO in in 9 CM O O C3 T- C3 C> O

d d d d d C3 C3

O O O O O O C D O

c o o 5 c o c M h - h - o 5 a 5 i n

d d d d d d d d d

O O C 3 O O O O O O ( 3 O

h - o c M c o o 5 i n c o i n i n c o m c o

C 3 d d d d d d c 3 d d c 3 c 3

C O O C O C O O r ; C O C O C M r ^ C M C O O

dddddddddoddd

00 O) 00 O 5 c q c 0 T - c 0 p C M C M C M CO CMi n c o " - ; S c n i - O 5 C M l J c M f , , .„ _ , .

c n T - c o - t C M c o c M r j c o S S c n o O C O 0 0 5 t ^ C O » - ' T C O . . C O C O CM r * . c o m o o o c o

o CM

. o i n o (O o

CO CM

(0 m a

CO t : t :

: ! ^ o

o a UJ

oCO

o d Q < O Q lU

CO CO CM

O o z

T N

60 DEVARAJ AND KOHLI

Table 6. Estimation Results for Profitability

Reference

Proposition 1

Proposition 3

Proposition 4

Control variabies

Variable

ITLAB0R(3)

ITSUPP(3) TCAP(4)

BPR(2)

ITLABOR(3)*BPR(2) ITSUPP(3)*BPR(2) ITCAP(4)*BPR(2)

FTE CASEMIX MEDICARE MEDICAID INCOME OUTPATNT

Model R-square

Model 1 Dependent

variable NPRDAY

0.005*

0.003

o.oor* 0.001

-0.001 0.001** 0.001 *

-0.429*** 233.29 619.75*

1978.00* -0.001

0.001

0.485

Model 2 Dependent

variable NPRADM

0.027**

0.010

0.031**

-0.006

0.009 0.031** 0.006*

- 2 . 1 6 * * * 2686.48

12973*** 11080**

0.263*** -0.015

0.795

a. The numbers in the table represent coefficients of independent variables. b. The number in parentheses indicates the number of lags in months of the independent variable. c. All performance data have been disguised by multiplying with a constant. *** indicates significance at the 1% level; ** indicates significance at the 5% level; * indicates significance at the 10% level.

patient revenue per day and net patient revenue per admission three periods later (at the 10 percent and 5 percent levels, respectively). Also, investments in IT capital have an effect on both net patient revenue per day and net patient revenue per admission four periods later (at the 5 percent level). However, we found no evidence ofthe relationship between IT support activities and profitability.

An examination ofthe coefficients associated with IT labor, support, and capital on measures of quality is presented in models 3 and 4 (presented in Table 7). IT labor investments in any given period significantly (at the 1 percent level) affect patient mortality three periods later. There was no relationship between IT capital and IT support investments and patient mortality. When the dependent variable is patient satisfaction, the only significant IT component is IT capital (at the 10 percent level) with a four-period lag.

In proposition 3, we articulated the relationship between BPR and organizational performance. Our data analyses suggest that BPR implementations affected both inpatient mortality and patient satisfaction as indicated by models 3 and 4. This is not to suggest that BPR does not have an impact on fmancial measures of hospital performance but that the combination of BPR and IT is what really affects fmancial performance.

IT PAYOFF IN THE HEALTH-CARE INDUSTRY 61

Table 7. Estimation Results for Quality Indicators MOD)

Reference

Proposition 2

Proposition 3

Proposition 4

Control variables

Variable

ITLAB0R(3) TSUPP(3) ITCAP(4)

BPR(2)

ITLABOR(3)*BPR(2) ITSUPP(3)*BPR(2) rrCAP(4)*BPR(2)

FTE CASEMIX MEDICARE MEDICAID

INCOME OUTPATNT

Model R-square

Model 3 Dependent

variable MORT

-0.158*** 0.001

-0.004

- ^ . 4 3 9 *

0.006 0.001 0.001

- 4 . 0 4 0 * 19468***

-21714*** 1559

0.163* 0.157

0.537

Model 4 Dependent

variable SAns

-0.006 0.001 0.003*

0.047**

0.022* 0.042* 0.001

0.001 - 2 8 8 . 1 7 * 1488.88*** 2720.78***

-0.001 0.011 *

0.513

a. The numbers in the table represent coefficients of independent variables. b. The number in parentheses indicates the number of lags in months of the independent variable. c. All performance data have been disguised by multiplying with a constant. * • • indicates significance at the 1% level; *• indicates significance at the 5% level; * indicates significance at the 10% level.

The combined effect of BPR and investment in IT can be examined by studying the sign and significance of the interaction terms for BPR and each of the IT components included in this study. Models 1 and 2 indicate strong support for the notion of fit between BPR and IT support, and BPR and IT capital. We found a positive and statistically significant relationship between these interaction terms and net patient revenue per day and net patient revenue per admission. When the dependent vari- ables are indicators of quality performance, the interaction terms that are statistically significant (at the 10 percent level) are the interaction between BPR and IT support and IT labor on patient satisfaction.

In the estimated models for measures of profitability, the nimiber of full-time em- ployees is significantly negatively associated with performance. Medicare and Med- icaid is positively associated with performance, and income is positively related to net patient revenue per admission. In the models for measures of quality, significant independent variables include case mix. Medicare, Medicaid, patient income, and outpatients. These independent variables are included in the study primarily as con- trol variables and are not the focus of our study. Therefore, for the sake of brevity, we do not discuss detailed implications of these findings.

Given the paucity of a priori theory on time lags in IT-performance studies, we

62 DEVARAJ AND KOHLI

conducted a "holdout" sample analysis to test the robustness of our results; the data set was split in two: a "holdouf sample and a test sample. Estimation analyses performed on these two sample groups indicated similar qualitative results (in terms of sign and significance of coefficients). Although our analyses employed the antici- pated cost savings from the various BPR projects as a measure of BPR implementa- tion, the results were similar to models that used a count of BPR initiatives. The robustness of the results to the various estimation methods is reassuring and lends credibility to our fmdings.

Conclusions

Significance of Investing in IT

THE RESULTS OF OUR STUDY INDICATE THAT INVESTING IN IT does lead to organiza- tional profitability. Because of our longitudinal research design, we were able to see the lag effects of such investment. We found that the profitability impact is seen in three months or more. We found a stronger IT-profitability link when the patient/ customer's overall profitability was considered, as opposed to per day of stay. This is probably due to the sporadic nature of treatment regimens provided during the pa- tients' stay. We found varying support for investment in IT labor and capital. How- ever, IT investment in constilting and other support services does not appear to have a direct impact on the profitability ofthe organizations. These fmdings are similar to the firm-level findings, as in Table 2, in other industries.

We also found support for IT investment's impact on quality initiatives within the organization. We, therefore, found support for propositions 1 and 2 that, in due course, IT investment leads to increased profitability and improved quality of prod- ucts and services.

Significance of BPR

The effectiveness of BPR initiatives has been a subject of discussion and controversy within organizations. We fmd that BPR initiatives lead to reduced mortality and increased patient satisfaction within organizations. As is intuitive, improvements resulting from BPR initiatives do not manifest immediately. Along these lines, we found that improvements in mortality and satisfaction resulting from BPR initiatives in organizations were realized after a period of two months. We did not fmd any evidence that BPR alone leads to improvement in profitability.

Significance of BPR with IT Investment

BPR often leads to turmoil within organizations because it demands modifications in how individuals and work processes operate. Therefore, the management and support

IT PAYOFF IN THE HEALTH-CARE INDUSTRY 63

of such change are critical to the success of BPR. In examining the combined effect of BPR and IT investment, we found strong support that it leads to improved profit- ability for the organization. Specifically, we found that IT support and BPR com- bined have the strongest impact on the two profitability measures. While no impact was found with BPR and IT labor investment, we did find evidence that IT capital investment combined with BPR affects profitability. These may represent using IT as an enabler for BPR such as a new information system to provide patient informa- tion at bedside. Such technologies have shown improved efficiency and effective- ness of patient care. These fmdings support the recent calls in the literature to study BPR effects in conjunction with the support environment within which BPR is implemented [5,47,62]. This may also explain the limited support found for propo- sition 3 (as above).

Limitations

This study employs data from hospitals belonging to a health system over a recent three-year period. Thus, a principal limitation of the study is the generalizability of the findings reported. Field studies, such as the one reported here, have the advantage of providing a richer operationalization of reality and the ability to track detailed data over time; however their generalizability to the larger population is limited. We do not mean to suggest that fmdings can be generalized to the larger population of hospitals, even though that may be a possibility. Neither can the fmdings be general- ized to other industries or other organizations.

Future Directions and Research

Large-sample, cross-sectional studies in conjunction with longitudinal studies are called for to examine the IT payoff issue in detail and to be able to make generaliza- tions across industries, firms, and the like. The random selection of companies through a survey-based approach would allow generalizations across the population of in- dustries or firms under consideration.

The literature in BPR implementation is rife with anecdotal evidence and short on rigorous empirical evidence of performance impacts of BPR. There is a defmite need to better measure BPR implementations through objective measures and to relate BPR to organizational performance in the context of other variables that also affect performance.

A more specific area for future research in health care is the impact of IT invest- ments in labor, support, and capital on the continuum of health care. This study is a first step in that direction, but it only examined the impact of IT on acute-care ser- vices. Other areas that represent the continuum are ancillary units, freestanding labo- ratories, outpatient services, and so on.

Finally, an implicit assumption in this study, and in most IT payoff studies, is that investments in IT translate into more usage of the IT, and that this usage results in better organizational performance. The IT usage component is often assumed but is

64 DEVARAJ AND KOHLI

empirically untested. This may be an interesting and challenging question for future research endeavors.

NOTES

Acknowledgments: The authors would like to thank Hank Groot for his continued support ofthe project, Frank Piontek for sharing his experiences and providing suggestions for clinical re- search, and Don Irmiger, Linda Martin, Doug Anthony, and Diana Utterback for their assistance in obtaining the data.

1. We also computed a measure of competition in the maricet This measure, called Herfmdahl 's index in the health care literature, is computed as a function of tbe beds in the market and beds in a specific hospital. However, due to minimal variation across the various time periods, the inclusion of this variable presented estimation problems. Hence, we chose to drop this variable from further consideration.

REFERENCES

1. Ahituv, N., and Giladi, R. Business success and information technology: are they really related? Proceedings ofthe Seventh Annual Conference ofManagement IS, Tel Aviv University, Israel, 1993.

2. Anderson E.W.; Fomell, C ; Rust R.T. Customer satisfaction, productivity, and profitabil- ity: differences between goods and services. Marketing Science, 16,2 (1997), 129-145.

3. Antonelli, C. New information technology and the knowledge-based economy: the Italian evidence. Review of Industrial Organization. 12,4(\ 997), 593-607.

4. Baily, M.N. What has happened to productivity growth? Science, 234 (1986) 443-451. 5. Barua, A.;Lee, B.;and Whinston, A. The calculus of reengineering. Information Systems

Research, 7,4 (1996), 409-428. 6. Barua, A., and Lee B. The information technology productivity paradox revisited: a theo-

retical and empirical investigation in the manufacturing sector. InternationalJoumal of Flexible Manufacturing Systems, 9,2 (1997), 145-166.

7. Barua, A.; Kriebel, C ; and Mukhopadhyay, T. Information technologies and busi- ness value—an analytic and empirical investigation. Information Systems Research, 6,1 (1995), 3-23.

8. Bemdt, E.R., and Morrison, C.J. High-tech capital formation and economic performance in U.S. manufacturing industries: an exploratory analysis. Journal of Econometrics, 65, 1 (1995), 5M3.

9. Brynjolfsson, E. The productivity paradox of information technology. Communications of the ACM, 35 (December 1993), 66-77.

10. Brynjolfsson, E. Technology's true payoff. InformationWeek(Octoher 10,1994), 34-36. 11. Brynjolfsson, E., and Hitt, L.M. Beyond the productivity paradox. Communications ofthe

ACM, 41, 8 (August 1998), 49-55. 12. Brynjolfsson, E., and Yang, S. Infonnation technology and productivity: a review of

literature. Advances in Computers, .^i (1996), 179-214. 13. Butters, S., and Eom, S. Decision support systems in the health care industry. Journal of

Systems Management (June 1992), 28-31. 14. Cameron, K.S., and Quinn, R.E. Organizational paradox and transformation. In R.E Quinn

and K.S. Cameron (eds.), Paradox and Transformation: Toward a Theory of Change in Orga- nization and Management. Cambridge, MA: Ballinger, 1988, pp. 1-18.

15. Caron, J.; Jarvenpaa, S.; and Stoddard, D. Business reengineering at CIGNA Corporation: experiences and lessons learned from the first five years. MIS Quarterly, 18,3 {\ 994), 233-250.

16. Chang, C , and Tuckman, H. The profits of not-for-profit hospitals. Journal of Health Politics, Policy, and Law, 13 (Fall 1988), 547-564.

17. Dewan, S., and Min, C. The substitution of information technology for other factors of production: a firm level analysis. Management Science, -^J, 12 (1997), 1660-1675.

IT PAYOFF IN THE HEALTH-CARE INDUSTRY 65

18. Dickinson, J.C. (1987). From process to policy: a generic prescription for test over- Utilization in the emergency department. Family Practice Research Journal, 7, 1 (Fall 1987), 12-21.

19. Diewert, W.E., and Smith, A.M. Productivity measurement for a distribution firm. National Bureau of Economic Research Working Paper no. 4812,1994.

20. Dudley, L., and Lasserre, P. Infonnation as a substitute for inventories. European Eco- nomic Review, 31 (1989), 1-21.

21. Ford, R.; Bach, S.; and Fottler, M. Methods of measuring patient satisfaction in health care organizations. Health Care Management Review, 22,2 (1997), 74-89.

22. Forgionne, G., and Kohli, R. HMSS: a management support system for concurrent hospital decision making. Decision Support Systems, 16,3 (1996), 209-229.

23. Forgionne, G.; Gangopadhyay, A.; Klein, J.; and Eckhardt, R. Using electronic commerce to improve health care management. Proceedings of the 1998 Association oflnformation Systems, Baltimore, 1998, pp. 293-295.

24. Francalanci, C , and Galal, H. Infonnation technology and worker composition: determinants of productivity in the life insurance industry. MIS Quarterly, 22, 2 (1998), 227-241.

25. Friedman, B., and Shortell, S. The fmancial performance of selected investor-owned and not-for-profit hospitals before and after Medicare Prospective Payment. Health Services Re- search, 23 (\9m), 237-267.

26. Furse, D.; Burcham, M.; Rose, R.; and Oliver, R. Leveraging the value of customer satisfaction information. Journal of Health Care Marketing, 14,3 (1994), 16-20.

27. Gapenski, L.; Vogel, B.; and Langland-Orban, B. The determinants of'hospital profitability. Health and Hospital Administration, 38, 1 (1993), 63-80.

28. Grover, V.; Teng, J.; Segars, A.; and Fiedler, K. The influence of infonnation technology diffusion and business process change on perceived productivity: the IS executive's perspective. Information and Management, 34,3 (1998), 141-159.

29. Hendricks, K., and Singhal, V. Does implementing an effective TQM program actually improve operating performance? empirical evidence fVom firms that have won quality awards. Management Science, 43,9 {1997), 1258-1274.

30. Hitt, L., and Brynjolfsson, E. Productivity, business profitability, and consumer surplus: three different measures of infonnation technology value. MIS Quarterly, 20,2 (1995), 121-142.

31. Joiner, G.A.; Salisbury, D.; and Bollin, G.E. Utilizing quality assurance as a tool for reducing the risk of nosocomial ventilator-associated pneumonia. American Joumal of Medical Quality, 11,2 (Summer 1996), 100-103.

32. Jonscher, C. An economic study of the infonnation technology revolution. InT.J. Allen and M.S. Scott Morton (eds.). Information Technology and the Corporation of the 1990s: Research Studies. Oxford: Oxford University Press, 1994, pp. 5-42.

33. Jorgenson, D.W., and Stiroh, K. Computers and growth. Economics of Innovation and New Technology, 3 (1995), 295-316.

34. Keen, P. Value analysis: justifying decision support systems. MIS Quarterly, J, 1 (1981), 1-15.

35. Kelly, M. Productivity and information technology: the elusive connection. Management Science. 40,11 (1994), 1406-1425.

36. Kohli, R.; Kerns, B.; and Forgionne, G. The application of TQM in a hospital's casualty and pathology departments. International Joumal of Quality and Reliability Management, 12,9 (1995), 57-75.

37. Kohli, R., and Forgionne, G. A model to measure DSS effectiveness: theory and empirical analysis. Revue des Systemes de Decision, 7 (1) (1998), 87-110.

38. Kohli, R.; Tan, J.; Ziege, D.; Piontek, F ; and Groot, H. Integrating cost infonnation with health care decision support systems. Topics in Health Care Information Management, 20, 1 (1999), 80-95.

39. Koski, H. The implications of network use, production network extemalities and public networking programmes for firm's productivity. Research Policy, 28,4 (1999), 423-439.

40. Kraemer, K., and Dedrick, J. Payoffs from investment in information technology—lessons from the Asia-Pacific region. World Development, 22,12(1994), 1921-1931.

41. Langland-Orban, B.; Gapenski, L.C.; and Vogel, W.B. Differences in characteristics of

66 DEVARAJ AND KOHLI

hospitals with sustained high and sustained low profitability. Hospital and Health Services Administration, 41,3 (Fall 1996), 385-399.

42. Lehr, W., and Lichtenberg, F. Computer use and productivity growth in U.S. federal government agencies, 1987-92. Joumai of Industrial Economics, 46,2(1998), 257-279.

43. Mahmood, M., and Soon S. A comprehensive model for measuring the potential impact of information technology on organizational strategic variables. Decision Sciences, 22,4 (1991), 869-897.

44. Mahmood, M., and Mann, G.J. How information technology investments affect organiza- tional productivity and performance: a longitudinal study. Proceedings ofthe 1997 Information Resources Management Association Intemational Conference, 1997,187—191.

45. Milgrom, P., and Roberts, J. Complementarities and fit: strategy, structure, and organiza- tional change in manufacturing. Joumai of Accounting and Economics, 7P (1995), 179-208.

46. Moad, J. Reengineering report from the trenches. Datamation, 41,5 (1994), 36-40. 47. Mooney, J.; Gurbaxani, V.; and Kraemer, K. A process oriented framework for assessing

the business value of infonnation technology. Database Advances in Information Systems, 27,2 (1996), 68-81.

48. Morrison, C , and Bemdt, E. Assessing the productivity of information technology equip- ment in U.S. manufacturing industries. National Bureau of Economic Research Working Paper no. 3582,1991.

49. Mukhopadhyay, T.; Rajiv, S.; and Srinivasan, K. Information technology impact on process outpnt and quality. Management Science. 43,12(1997), 1645-1659.

50. Myers, B.; Kappelman, L.; and Prybutok, V. A comprehensive model for assessing the quality and productivity ofthe information system function: toward a contingency theory for infonnation systems assessment University of North Texas, Working paper, ISRC-WP-19970101, 1997.

51. Pinsonneault, A., and Rivard, S. Information technology and the nature of managerial work: from the productivity paradox to the Icarus paradox? MIS Quarterly, 22, 3 (1998), 287-311.

52. Prasad, B., and Harker, P. Examining the contribution of information technology toward productivity and profitability in U.S. retail banking. Working paper no. 97-09, Financial Institu- tions Center, The Wharton School. 1997.

53. Prattipati, S., and Mensah, M. Information systems variables and management productiv- ity. Information and Management, 33,1 (1997), 33-43.

54. Quinn, J. The productivity paradox is false: information technology improves services performance. Advances in Services Marketing and Management, 5 (1996), 71—84.

55. Roach, S. America's Technology Dilemma: A Profile ofthe Information Economy. Special Economic Study, Morgan Stanley, 1987.

56. Robey, D. Theories that explain contradiction: accounting for the contradictory organiza- tional consequences of infonnation technology. Proceedings ofthe Sixteenth Intemational Con- ference on Information Systems, Amsterdam, December 1995, pp. 55—63.

57. Rutledge, R. Life after the Deming Prize. Business and Economic Review, 40, 3 ( 1 9 9 4 ) 2 6 - 3 1 .

58. Seddon, P. A respecification and extension of the DeLone and McLean model of IS success. Information Systems Research, 8,3 (1997), 240-253.

59. Shi, L. Patient and hospital characteristics associated with average length of stay. Health Care Management Review, 21,2 (1996), 46-61.

60. Siegel, D., and Griliches, Z. Purchased services, outsourcing, computers, and productivity in manufacturing. In Z. Griliches et al. (eds.). Output Measurement in the Service Sectors. Chicago: University of Chicago Press. 1992.

61. Silagy, C , and Lancaster, T. The Cochrane Collaboration in Primary Care: an intemational resource for evidence-based practice of family medicine. Family Medicine, 27,5 (May 1995), 302-305.

62. Soh, C , and Markus, M. How IT creates business value: a process theory synthesis. Proceedings ofthe Sixteenth International Conference on Information Systems, Amsterdam, December 1995, pp. 29-41.

63. Strassman, P.A. The Business Value of Computers. New Canaan, CT: Information Eco- nomics Press, 1990.

IT PAYOFF IN THE HEALTH-CARE INDUSTRY 67

64. Strassmann, PA. Information Payoff: The Transformation of Work in the Electronic Age. New York: Free Press, 1985.

65. Tam, K. Analysis of firm-level computer investments: a comparative study of three Pacific- Rim economies. IEEE Transactions on Engineering Management, 45,3 (1998), 276-286.

66. Teisberg, E.; Porter, M.; and Brown, G. Making competition in health care work. Harvard Business Review, 72,4 (1994), 131-141.

67. Teo, T , and Wong, P. An empirical study of the performance impact of computerization in the retail industry. Omega: Intemational Joumal of Management Science, 26, 5 (1998), 6 1 1 - 621.

68. Teplensky, J.D.; Pauly, M.V.; Kimberly, J.R.; Hillman, A.L.; and Schwartz, J.S. Hospital adoption of medical technology: an empirical test of altemative models. Health Services Re- search, 30, 3 (1995), 437-465.

69. Van de Ven, A.H., and Drazin, R. The concept of fit in contingency theory. Research in Organizational Behavior, 7(1985), 333-365.

70. Venkatraman, N. The concept of fit in strategy research: toward verbal and statistical correspondence. Academy of Management Review, 14,3 (1989), 423-444.

71. Vogel, W.; Langland-Orban, B.; and Gapenski, L.C. Factors influencing high and low profitability among hospitals. Health Care Management Review, 18,2 (1993). 15-26.

72. Wachtel, T.J., and O'Sullivan, P. Practice guidelines to reduce testing in the hospital. Journal of General Internal Medicine, 5,4 (July-August 1990), 335-341.

73. Wilcocks, L., and Lester, S. In search of information technology productivity: assessment \s&\ii%. Journal of the Operational Research Society, 48,11 (1997), 1082-1094.

74. Wilson, D. Assessing the impact of information technology on organizational perfomiance. In R. Banker, R. Kauffhian, and M.A. Mahmood (eds.). Strategic Information Technology Management. Harrisburg, PA: Idea Group, 1993.