Feb 14, 2012 - is mitigated in hospitals with some internal information technology expertise,; ... costs two years after adoption basic EMR and no increase in costs two ...... and plausibility: Next, we explore the degree to which we can claim.
The Trillion Dollar Conundrum: Complementarities and Health Information Technology
February 2012 David Dranove (Northwestern University) Chris Forman (Georgia Institute of Technology) Avi Goldfarb (University of Toronto) Shane Greenstein (Northwestern University)
Abstract
We examine the relationship between the adoption of electronic medical records (EMR) and hospital operating costs at thousands of US hospitals between 1996 and 2008. We combine data from the American Hospital Association Annual Survey, the Healthcare Information and Management Systems Society Analytics Database, the US census, and Harte Hanks Market Intelligence information technology survey. We first identify a puzzle that has been seen in prior studies: Adoption of EMR is generally associated with a slight increase in costs. We argue that this average masks important differences over time, across locations, and across hospitals. Drawing on the literature on information technology as a business process innovation, we show that: (1) EMR adoption is initially associated with a rise in costs; (2) the initial increase in costs is mitigated in hospitals with some internal information technology expertise,; (3) Hospitals in locations that have local information technology expertise can experience a decrease in costs after two years; and (4) Hospitals in other locations experience a sharp increase in costs even after six years.
I. Introduction As annual U.S. healthcare expenditures climb towards $3 trillion (with spending forecast to exceed $4.5 trillion by 2020), many analysts hope that electronic medical records (EMR) can stem the tide (Centers for Medicare & Medicaid Services). Melinda Beeuwkes Buntin and David Cutler make EMR the centerpiece of their “Two Trillion Dollar” solution for modernizing the health care system (Buntin and Cutler 2009). Yet, the Congressional Budget Office states: “No aspect of health information technology entails as much uncertainty as the magnitude of its potential benefits” (Congressional Budget Office 2008).1 A small sampling of research from the last half dozen years provides a sense of the uncertainty. A widely cited 2005 report by the RAND Corporation, published in the leading policy journal Health Affairs, estimates that widespread adoption of EMR by hospitals and doctors could reduce annual health spending by as much as $81 billion while simultaneously leading to better outcomes (Hillestad et al. 2005). Jaan Sidorov, a medical director with the Geisinger Health Plan, an early adopter of EMR, published a response to the RAND report in Health Affairs. Sidorov (2006) notes the high cost of adoption and cites evidence that EMR leads to greater health spending and reduced provider productivity. Other recent studies, cited below, fail to find consistent evidence that EMR savings offset adoption costs. Thus, it remains uncertain whether EMR can fulfill its promise and bring about major reductions in health spending. This study speaks to this conundrum and reframes this debate. We characterize EMR in terms of the impact new information technology (IT) has on existing enterprises. Specifically, we view EMR as a business process innovation, which is a change in the operational practices inside the adopting organization. Like other business process innovations, the impact of new IT such as EMR depends on its complementary with local labor markets, on its complementary with internal expertise, and on the functional heterogeneity in the new process’ components. We conclude that it is premature to assess the impact of EMR without consideration of these three factors. We apply this reframing to statistical data, examining the impact of EMR adoption on hospital operating costs during the period 1996 to 2008. We demonstrate the importance of complementarities with local conditions and internal expertise by testing for differences across regional labor markets and across hospitals with different prior experiences with management technology. We test for functional heterogeneity by distinguishing between the introduction of the more basic and more advanced types of healthcare information technologies. We combine several datasets to link hospital costs to EMR adoption and the potential for complementarities.2 The data reveal a complex pattern of expenditures and savings that is remarkably consistent with the framing of the problem. We find that, on average, hospitals that adopted EMR between 1996 and 2008 did not experience a statistically significant increase in operating efficiency. In fact, under many specifications costs rose after EMR adoption. However, this effect is mediated by the availability of technology skills in the local labor market. Specifically, in strong IT locations costs fall sharply after the first year of adoption to below pre1
The term Health Information Technology is often used interchangeably with EMR. We use data from the American Hospital Association Annual Survey (hospital characteristics), from the medical costs report (hospital costs), from the Healthcare Information and Management Systems Society Analytics Database (hospital EMR adoption), from the US census (local demographics and industry), and from Harte Hanks Market Intelligence information technology survey (local business IT, hospital IT capabilities in 1996). 2
2
adoption levels. In weak IT locations, costs remain above pre-adoption levels indefinitely. Overall, hospitals in high skill markets enjoyed a statistically significant four percent decrease in costs two years after adoption basic EMR and no increase in costs two years after adoption of advanced EMR. These are significantly lower than the two to three percent rise in costs after adoption by hospitals in other markets. Complementary skills can also be found in the hospital itself. For advanced EMR, the initial increase in costs is mitigated substantially if hospitals already employ a programmer or if hospitals have used a clinical software application prior to adoption. Figure 1 displays these general patterns, comparing hospitals that adopt basic and advanced EMR before the adoption period, during the adoption period, and after the adoption period. For basic EMR, costs do not fall until two years after adoption. For non-IT intensive locations, costs rise sharply in the year of adoption, and then fall back. For IT intensive locations, costs fall with adoption, and are substantially lower two years after adoption. Hospitals with programmers by 1996 display a similar pattern (though the results on basic EMR adoption are not significant). For advanced EMR, the patterns are similar but sharper: costs rise in the period of adoption for non IT intensive locations and hospitals without programmers and costs fall over time for the other hospitals. By providing an explanation for the mixed results of past studies, our analysis informs the understanding of EMR’s sluggish diffusion. As of 2009, only about 60-70 percent of America’s hospitals have adopted key elements of EMR.3 So sluggish was this adoption that it motivated public intervention. In order to spur EMR adoption, Congress in 2009 passed the Health Information Technology for Economic and Clinical Health Act (EMRECH Act), which provides $20 billion in subsidies for providers who adopt EMR. Two thirds of hospitals say they plan to enroll in the first stage of EMRECH subsidy programs by the end of 2012 (US Department of Health and Human Services 2011). In addition, the 2010 Patient Protection and Affordable Care Act promote EMR adoption. PPACA directs the establishment of quality reporting measures that likely will require providers to adopt EMR in order to comply. PPACA creates a new “shared savings” program for Medicare; participation in this program is predicated on the use of EMR. Finally PPACA encourages providers to apply to participate in a range of new programs and gives preference to those that have adopted EMR. Most importantly, our findings may help resolve the ongoing debate between supporters and detractors of EMR. Both sides have seemed to treat EMR as if its economic impact is independent of other environmental factors, as if it either works or it doesn’t. This creates a conundrum for both sides. If EMR is going to save as much as a trillion dollars, as its supporters claim, why isn’t it working in obvious ways? If it costs more than it saves, as its detractors argue, why are policy makers so keen to expand adoption? Our results suggest that the debate about EMR can be resolved by drawing on the general literature on business process innovation, where it is very common for successful adoption of enterprise IT to require complementary changes in business processes and related labor inputs. It is also common for new enterprise IT to be more productive when companies have access to these inputs in their local labor market. Using this experience, it is not surprising that EMR can simultaneously have the potential to generate substantial savings yet demonstrate mixed results in practice. 3
Source: Authors’ calculations based on data supplied by HIMSS.
3
Our findings therefore provide a possible resolution to the trillion dollar conundrum. EMR can succeed when the necessary complementarities are present and the complementary components are in place. Until then the results, at best, can be only mixed. Over time, these complementarities are expected to become more widely available, and the various components more widely deployed. If so, EMR may yet fulfill its promise. While EMR’s past mixed performance is no guarantee of a future result, the past experience also is no guarantee of future failure. As these skills spread, hospitals in other markets may eventually enjoy the same benefits. We proceed as follows. Sections II and III describe the institutional setting for EMR, and some of the prior evidence about its effects on hospitals. This motivates a comparison in Section IV between EMR and the adoption of IT inside organizations. That leads to a reframing of several key hypotheses. Sections V and VI present data and results. Section VII concludes. II. What is EMR? It is essential to define terms. EMR is a catchall expression used to characterize a wide range of information technologies used by hospitals to keep track of utilization, costs, outcomes, and billings. In practice, EMR includes, but is not limited to:
A Clinical Data Repository (CDR) is a real time database that combines disparate information about patients into a single file. This information may include test results, drug utilization, pathology reports, patient demographics, and discharge summaries Clinical Decision Support Systems (CDSS) use clinical information to help providers diagnose patients and develop treatment plans Order Entry provides electronic forms to streamline hospital operations (replacing faxes and paper forms). Computerized Physician Order Entry (CPOE) is a more sophisticated type of electronic order entry and involves physician entry of orders into the computer network to medical staff and to departments such as pharmacy or radiology. CPOE systems typically include patient information and clinical guidelines, and can flag potential adverse drug reactions. Physician Documentation helps physicians use clinical information to generate diagnostic codes that are meaningful for other practitioners and valid for reimbursement
As this list shows, there is no one single technology associated with EMR, and different EMR technologies may perform overlapping tasks. Our data from HIMSS Analytics contain biannual hospital-level adoption data for each of these technologies. Therefore, we are able to explore how different technologies might affect costs in different ways. Nearly all of the information collected by EMR resides in hospital billing and medical records departments and in physicians’ offices. EMR automates the collection and reporting of this information, including all diagnostic information, test results, and services and medications received by the patient. EMR can also link this information to administrative data such as insurance information, billing, and basic demographics. EMR can reduce the costs and improve the accuracy of this data collection. Two components of EMR, Clinical Decision Support Systems and Computerized Physician Order Entry, use clinical data to support clinical decision making. If implemented in ideal conditions and executed according to the highest standards,
4
EMR can reduce personnel costs while facilitating more accurate diagnoses, fewer unnecessary and duplicative tests, and superior outcomes with fewer costly complications. Despite these potential savings, EMR adoption has been uneven. Table 1 reports hospital adoption rates for the five components of EMR described above. The data is taken from HIMSS Analytics, which we describe in more detail in section V. Clinical Data Repository, Clinical Decision Support, and Order Entry are older technologies that were present in many hospitals in the 1990s. Even for these older technologies, adoption rates were below 85 percent in 2008. The remaining applications emerge in the early to mid-2000s. Adoption rates for these are below 25 percent. While informative, Table 1 lacks several crucial pieces of information. It lacks comparable data on physician adoption of EMR, for example. The conventional wisdom is that physician adoption rates are much lower. In addition, adoption of such systems provides no information about intensity of use, or on how close operations come to ideal conditions. Once again, conventional wisdom suggests that many hospitals have experienced a wide range of outcomes, and in some cases this is due to poorly executed installations, or lack of ideal conditions for hiring skilled talent. Although beyond the scope of this study, compatibility issues also may shape the success of EMR at a regional level, and this too is missing from the table. There are many different EMR vendors and their systems do not easily interoperate. As a result, independent providers cannot always exchange information, which defeats some of the purpose of EMR adoption (Miller and Tucker 2009). The EMRECH Act widens the scope of privacy and security protections and may therefore make it easier for different vendor systems to exchange information in the future. III. Evidence on the Potential Savings from EMR Has the adoption of EMR yielded gains to prior adopters? This section reviews prior evidence, stressing the absence of work focusing on operational savings, lack of emphasis on complementarities with the labor market, and the absence of accounting for the functional heterogeneity of EMR’s components. This discussion will motivate our concern with these aspects. EMR is expensive. One prominent estimate, from the Congressional Budget Office, estimates that the cost of adopting EMR for office-based physicians lay between $25,000 and $45,000 per physician, with annual maintenance costs of $3000 to $9000. For a typical urban hospital, these figures range from $3-9 million for adoption and $700,000-$1.35 million for maintenance. In context these costs are quite significant: If the adoption costs are amortized over ten years, EMR can account for about 1 percent of total provider costs. It would be no surprise, therefore, if available research suggests that EMR may not pay for itself, let alone generate hundreds of millions of dollars in savings. In their review of 257 studies of EMR effectiveness, Chaudry et al. (2006) note that most studies to date do not focus on cost savings, providing at best just indirect evidence of productivity
5
gains.4 Most of the studies they review focus on quality of care. Ten studies examine the effects of EMR on utilization of various services. Eight studies show significant reductions of 8.5-24 percent, mainly in laboratory and radiology testing. While fifteen studies contained some data on costs, none offered reliable estimates of cost savings. Indeed, only three reported the costs of implementing EMR and two of these studies were more than years old. One of the most widely cited studies about costs savings, Hillestad et al. (2005) (the RAND study cited in our introduction), uses results from studies of EMR and medical utilization up until 2005, and extrapolates the potential cost savings net of adoption costs. They identify several dozen potential areas of cost savings, including reduced drug, radiology, and laboratory usage, reduced nursing time, reductions in clerical staff, fewer medical errors, and shorter inpatient lengths of stay. They estimate that if 90 percent of U.S. hospitals were to adopt EMR, total savings in the first year would equal $41.8 billion, rising to $77.4 billion after fifteen years. They also predict that EMR adoption could eliminate several million adverse drug events annually, and save tens of thousands of lives annually through improve chronic disease management. Sidorov (2006) challenges these findings, arguing that the projected savings are based on unrealistic assumptions. For example, the RAND study appears to assume that EMR would entirely replace a physician’s clerical staff. It is well-known, however, that providers who adopt EMR tend to reassign staff rather than replace them. To take another example, EMR is supposed to eliminate duplicate tests, while it is just as likely that, in reality, EMR may allow providers to justify ordering additional tests. Sidorov also questions whether EMR will generate forecasted reductions in medical errors. Enough time has passed to compare RAND’s forecasts against actual experience. The most recent review of the literature on EMR effectiveness is by Buntin et al. (2011). They identify 73 studies of EMR and medical utilization. EMR is associated with a significant reduction in utilization in 51 (70 percent) of these studies. They do not break these down into specific areas of savings, however. Moreover, Buntin et al. do not identify any studies of EMR and costs. To our knowledge, such studies remain few and far between. Indeed, we have identified only three focused cost study. Borzokowski (2009) uses fixed effects regression to examine whether early versions of financial and clinical information technology systems generated significant savings between 1987 and 1994. He finds that the most thoroughly automated hospitals generate up to 5 percent savings within five years of adoption. He also finds that hospitals that adopt EMR but are not the most thoroughly automated experience an increase in costs. In this way, his conclusions mirror the popular discussion: there appears to be the potential for savings but there is little understanding of the drivers of the heterogeneity across hospitals. Furukawa, Raghu, and Shao (2010) study the effect of EMR adoption on overall costs among hospitals in California for the period 1998-2007. Using fixed effects regression, they find that EMR adoption is associated with 6-10 percent higher costs per discharge in medical-surgical acute units, in large part because nursing hours per patient day increased by 15-26 percent. This is plausible because nurse use of EMR can be very time consuming. Finally, Agha (2012) uses 4
Chaudry et al state that they are studying Health Information Technology and they do not indicate if they distinguish between HIT and EMR.
6
variation in hospitals’ adoption status over time, analyzing 2.5 million inpatient admissions across 3900 hospitals between the years, 1998-2005. Health IT is associated with an initial 1.3 percent increase in billed charges. She finds no evidence of cost savings, even five years after adoption. Additionally, adoption appears to have little impact on the quality of care, measured by patient mortality, medical complication rates, adverse drug events, and readmission rates. None of the studies to date frame EMR as a business process innovation. In other words, there is no effort to investigate whether the impact of EMR might vary systematically with the type of EMR adopted, or with the factors that shape availability of complementary components, such as the characteristics of the local settings. This may be due to the absence of a theory that would suggest such differential effects. In the next section, we offer such a theory. IV. Information Technology and Complementarities Business process innovations alter organizational practices, generally with the intent of improving services, reducing operational costs, and taking advantage of new opportunities to match new services to new operational practices. Typically this type of innovation involves changes in the discretion given to employees, changes to the knowledge and information that employees are expected to retain and employ, and changes to the patterns of communications between employees and administrators within an organization. Because important business process innovations occur on a large scale, they typically involve a range of investments, both in computing hardware and software, and in communications hardware and software. They also involve retraining of employees, as well as redesign of organizational architecture, such as its hierarchy, lines of control, compensation patterns and oversight norms. Large scale business process innovations do not necessarily display patterns observed in familiar and standard productivity models for new product or process innovations. Business process innovations within enterprises do not necessarily follow a single diffusion curve, as typically found in a monolithic new product or process replacing an older one. Related, the first investments associated with adoption do not necessarily and reliably generate a productivity impact. This is because the sequence of complementary investments and organizational changes after the initial adoption play such an important role in determining the eventual productivity of the changes. By itself new IT has little impact. Prior studies place stress the importance of co-invention, the post-adoption invention of complementary business processes and adaptations aimed at making the adoption useful (Bresnahan and Greenstein, 1996). That is because IT often is affiliated with other infrastructure investments at the organizational level, which enables change, but it is not sufficient for ensuring that the change occurs. The latter depends on whether the employees of the adopting organization – in the case of hospitals, administrative staff, doctors, and nurses – find new uses and invents new services. The latter may depend on the conditions of the local labor market for skilled talent. It can also depend on supply conditions, particularly those practices which shape the ease with which lessons from experiments in one organization becomes widely known. For example, do suppliers or third-party consultants quickly transmit lessons across adopting organizations? In the absence of such practices, high payoff to early adoption may face considerable delays.
7
Prior studies also place importance on the time it takes to implement change in large organizations. Interruptions to ongoing operations generate large opportunity costs in foregone services. Such factors may slow down implementation or the realization of returns, breaking the link between the order of initial adoption and the extent of the payoff. Indeed, the earliest adopters may face the largest transition costs because society has not yet gained economies of scale in learning. Therefore, prior studies stress that business process innovations usually do not yield immediate payoffs to adopting organizations (Bresnahan, Brynjolfsson, Hitt, 2002). The importance of complementary factors and co-invention leads to considerable variance in postadoption utilization and further investment. Hence, the incentives around utilization and investment can change considerably over time due to changes in the supply of complements and differences in the restructuring of organization’s hierarchy and operational practices. There is often little immediate payoff to adoption, and a strong potential for lagged payoff, if any arises at all. Where would these factors be most visible? For the most recent vintages of information technology, evidence suggests considerable heterogeneity across US locations in the availability of complementary factors, such as skilled labor (Forman, Goldfarb, and Greenstein, 2005, 2012), third-party software support and service (Arora and Forman, 2007), and infrastructure (Greenstein and McDevitt, 2011). That induces a visible relationship between investment in health IT and geographic location. Large cities may have thicker labor markets for complementary services or for specialized skills. Thicker markets lower the (quality-adjusted) price of obtaining IT services such as contract programming and of hiring workers to perform development activities in-house. Such locations may also have greater availability of complementary information technology infrastructure, such as broadband services. Increases in each of these factors may decrease the costs of adopting complex technologies in some cities and not others, other things being equal. Overall, the presence of thicker labor markets for technical talent, greater input sharing of complex IT processes, and greater knowledge spillovers in cities should lower the costs of successful adoption of frontier technologies in big cities (Henderson 2003; Forman, Goldfarb, Greenstein 2008). We also expect enterprises with existing IT facilities will be more likely to adopt frontier IT than establishments without an extensive operations. Having more resources elsewhere in the organization means that lower cost resources can be tapped, or loaned between projects of the same firm. Also, the internal firm resources that arise as a result of prior investments in other IT projects may lower adoption costs. Resources and other investments in the organization are already employed in some IT task, and the new technical opportunity leads them to be redeployed for use in advanced Internet applications. Programmers provide experience with IT projects. Prior IT projects may reduce development costs if programmers are able to transfer lessons learned from one project to another.5 Prior work on other IT projects may create learning economies and spillovers that decrease the costs of adapting general purpose IT to organizational needs, reducing the importance of external consultants and local spillovers. For example, Forman, Goldfarb, and Greenstein (2008) documented that, when IT labor forces are mobile, 5
For example, software developers may be able to share common tools for design, development, and testing (Banker and Slaughter, 1997), or they may be able to reuse code (Barnes and Bollinger, 1991). Software development may also have learning economies (Attewell, 1992) that through experience reduce the unit costs of new IT projects. Much prior research in the costs of innovative activity has also long presumed experience with prior related projects can lower the costs of innovation (Cohen and Levinthal, 1990).
8
shared human capital at other establishments decreases the value of consultants and thicker labor markets in large cities. This should shape hospital investment due to use of statewide and federal funding mechanisms. Therefore, we expect less connection between local characteristics and the impact of advanced IT at hospitals with prior IT experience because such organizations would have more resources for overcoming the limitations of their locations. We would add an additional observation, namely, we expect the prevalence of the various local factors that generate complementarities (IT-intensive industries, high population, etc.). This is due to the virtuous cycles – i.e., positive reinforcement – that develop between these factors. For example, a hospital in an urban location may be able to take advantage of a frontier component of EMR due to a thick labor market in a major city, tapping local technical expertise, while one in a suburban location may not have such options. V. Data We use a variety of data sources to examine the relationship between EMR adoption and costs. In particular, we match data on EMR adoption from a well-known private data source on health IT investments (HIMSS Analytics) with cost data from the Medicare Hospital Cost Report. We add to this data from the American Hospital Association’s (AHA) Annual Survey of Hospitals. We obtain regional controls and information on local complementary factors from the decennial U.S. Census and from U.S. County Business Patterns data. We supplement the sources above with information on lagged hospital-level IT capabilities from another private source on IT investment, the Harte Hanks Computer Intelligence Database. Our data are organized as an unbalanced panel, with data available for every other year over 1996 – 2008. Table 2 provides descriptive statistics.6 EMR data. We obtain information about EMR adoption from the Healthcare Information and Management Systems Society (HIMSS) Analytics data base. The HIMMS Annual Study collects information systems data related to software and hardware inventory and relates the current status of EMR implementation of more than 5300 healthcare providers nationwide, including about 2000 hospitals. Organizations that seek access to HIMSS Analytics data must provide their information on software and hardware use. Because most organizations tend to participate for a long period of time, the HIMSS Analytics data closely approximates panel data and can be used for fixed effects regression. HIMSS reports adoption of 99 different technologies in 18 categories. Examples include Emergency Department Information System in the Emergency Department; Financial Modeling for Financial Decision Support, and a Laboratory Information System for the Laboratory. Following most other studies, we restrict attention to six applications in the category Electronic Medical Records, which we listed above. These come closest to representing the kind of EMR applications that the RAND study and others believe will lead to dramatic cost savings and quality enhancements. We also examine two additional margins of adoption that aggregate these six technologies into broad categories of usage that we label basic and advanced EMR adoption. Applications within each of these categories will involve similar costs of adoption and require 6
The number of observations column in table 1 shows a key challenge within and across data sources: missing data. There is considerable variation across hospitals and years for each of the variables. We simply drop missing observations from our main specifications, but results are robust to alternatives.
9
similar types of co-invention to be used successfully. The variable basic EMR is equal to one if the hospital adopts a clinical data repository (CDR), clinical decision support systems (CDS), or order entry/communication. The variable advanced health IT is equal to one when the hospital adopts either computerized practitioner order entry (CPOE) or physician documentation, applications which are more difficult to implement and, for example, are often included in the latter stages of EMR adoption in analyses of health IT adoption like the HIMSS Forecasting Model (HIMSS Analytics 2011). Our estimation sample is based on the set of hospitals who replied to the HIMSS survey. Thus, we may exclude hospitals who systematically invest little in information systems and so have little incentive to reply to the HIMSS survey. Table 2 shows that, by 2008, at least 70 percent of responding hospitals had adopted each of the basic EMR technologies and at least 20 percent had adopted the advanced technologies. The bottom of the table shows the changes from 1996 and documents sharp increases in adoption of all of these technologies over the sample period. Cost Data. We collect data on hospital costs from the Medicare Cost Report. Our primary dependent variable is equal to total operating expenses at the hospital, but we also examine specific cost categories that may be particularly affected by EMR, including total diagnostic radiology costs and total nursing administration costs. The bottom of table 2 shows that, on average, costs rise considerably over the sample period but there is a great deal of variation across hospitals. AHA Data. We obtain hospital characteristics from the American Hospital Association Annual Survey. The survey contains details about hospital ownership, service offerings, and financials. We match AHA, Cost Report, and HIMSS data using the hospital Medicare ID and retain only matching hospitals. Our final data set contains 4413 hospitals, 97 percent of which are observed in all eight years of the data. Missing data about specific technologies and about hospital characteristics mean that our regressions involve 2596 to 3515 hospitals observed an average of 4 to 5 periods. We use information from the AHA data and the Medicare Cost Report to exclude several types of hospitals whose costs might be affected by unobservable and/or idiosyncratic factors unrelated to EMR adoption. In particular, we exclude federal hospitals, as well as hospitals that are not defined as short-term general medical and surgical hospitals. (The hospitals that we exclude are not usually considered to be “community” hospitals). Finally, we dropped a small number of hospitals that report very low total costs (less than $1000) over one or more years over our sample period. After dropping these, the minimum cost is $467,530, the average cost is $108 million and 99% of the data have costs above $2.6 million. We use the AHA data to compute the following key predictors:
Hospital Size: We include total number of hospitals beds, number of admissions, number of inpatient days, number of full-time physicians and dentists, and the number of discharges (Medicare, Medicaid, and total).
10
Hospital Organization: We include indicators of whether the hospital is an independent practice association hospital, or a management service organization hospital. We also include two indicators for whether the hospital is vertically integrated. The first measure is equal to one when the hospital is part of an open hospital physician organization, a closed hospital physician organization, an integrated salary model, or part of a group practice without walls. The second measure is identical but excludes the integrated salary model from the definition of vertical integration in order to allow separate influences on costs for these types of vertical integration. Service Composition: Including number of births, total number of outpatient visits, and percent births. Hospital ownership: Including indicators of for-profit ownership, non-secular nonprofit ownership, non-profit church ownership, an equity model hospital, or a foundation hospital. Other characteristics: Including whether the hospital is a teaching hospital (defined as having a residency program or being a member of the council of teaching hospitals), and the number of physicians divided by the total number of hospital beds.
Census Data. We use U.S. Census data to identify location-level factors that might affect costs independent of information technology and to measure complementary factors that might facilitate process innovation. For controls, we obtain the following variables from the 2000 decennial U.S. Census and match on county: population, percent Hispanic and Black, percent age 65+ and percent age 25-64, percent university and high school education, median home value, and median household income. To measure the availability of local complementary factors, we use two measures from the Census. First, we include a dummy for whether the hospital is located in an MSA. Urban locations will benefit from additional supply of complementary factors, including thicker labor markets, third party services firms, and better infrastructure.7 Urban location has been shown to be correlated with frontier IT adoption in a variety of settings (Forman, Goldfarb, and Greenstein 2005, 2008). Our second measure of complementary factors is the percentage of local firms that are in ITusing industries. Like location in an MSA, we use this measure to capture geographic variance in the demand and supply for complementary factors that may affect the returns to EMR adoption. We measure the fraction of firms in IT-using and IT-producing industries in the county as of 1995 from the US Census County Business Patterns data. National aggregate data shows that such industries have unusually high returns from investment in IT in the 1990s. We define these industries using the classification reported in Jorgenson, Ho, and Stiroh (2005, p. 93).8 Table 2 7
Of course, urban location will also have stronger demand for the same factor, making identification of the relationship between IT investment, urban location, and hospital costs difficult. 8 These industries are Communications (SIC 48), Business Services (73), Wholesales Trade (50-51), Finance (60-62, 67), Printing and Publishing (27), Legal Services (81), Instruments and Miscellaneous Manufacturing (38-39), Insurance (63-64), Industrial Machinery and Computing Equipment (35), Gas Utilities (492, 496, and parts of 493), Professional and Social Services (832-839), Other Transportation Equipment (372-379), Other Electrical Machinery (36, ex. 366-267), Communications Equipment (SIC 366), and Electronic Components (367).
11
shows that 43 percent of the hospitals in our data are in counties in the top quartile in IT intensity. Additional IT Data. To obtain measures of historical hospital-level IT capabilities, we gather data from the Harte Hanks Market Intelligence Compute Intelligence Technology Database (hereafter CI database). The CI database contains establishment- and firm-level data on characteristics such as the number of employees, personal computers per employee, number of programmers, and the use of specific software applications. A number of researchers have used this data previously to study adoption of IT (e.g., Bresnahan and Greenstein 1996) and the productivity implications of IT investment (e.g., Bresnahan, Brynjolfsson, and Hitt 2002; Brynjolfsson and Hitt 2003; Bloom et al. 2009). Interview teams survey establishments throughout the calendar year; our main sample contains the most current information as of December 1998. As has been discussed elsewhere (e.g., Forman, Goldfarb, and Greenstein 2005), this data set represents among the best sources of information on the IT investments of private firms available. We use the CI database to obtain measures of lagged IT capabilities of hospitals and measures of IT-intensive location. For capabilities of hospitals, we gather data on the number of computer programmers at the hospital and whether the hospital had at least one clinical application in 1996. We merge information from the CI database using hospital names. Unfortunately, because the CI database is itself a sample from a broader population of firms, there is a significant loss of data from merging these two data sources: the number of hospitals in our sample falls by more than half in the regressions that use the CI database directly to measure whether the hospital has programmers or clinical applications by 1996. 20 percent of the hospitals in this smaller sample had at least one programmer on staff in 1996 and 67 percent of these hospitals used at least one clinical software application. For measures of IT-intensive location, we rely on calculations in Forman, Goldfarb, and Greenstein (2012) to identify the percentage of firms in a county that had adopted basic and advanced internet technology by 2000. These calculations use the most current information in the CI database as of December 2000. They represent the average level of internet adoption by firms in the county, weighted so that the Harte Hanks database represents the distribution of firms in the industry as identified in the economic census. Basic internet technology includes simple applications such as web browsing and email. Advanced internet technology is identified by ecommerce or e-business applications that involve frontier technologies and significant adaptation costs. We do not emphasize these measures of local IT-intensity, but use them to show that one particular definition of IT-intensive location does not drive our results. V. Empirical strategy and results We perform linear regression with hospital and year fixed effects on an unbalanced panel of hospitals observed on even years from 1996 to 2008. We proceed in four stages. First, we regress costs on different measures of EMR adoption and document that costs appear to rise on average after adoption. Second, we decompose the rise in costs by years since adoption and show that the rise is largest in the first year of adoption. Third, we examine different margins of complementarity: location and internal IT experience. Finally, we examine robustness, identification, and plausibility with a variety of further tests. 12
Overall effects: We begin by examining the relationship between total administrative costs and EMR: (1)
Log(cit) =Xit+tZi+EMRit+t+i+it,
Here, t is a time dummy that captures average changes to costs over time; i is a hospitalspecific fixed effect that gets differenced out in the estimation; and EMRit is a discrete variable for whether hospital i had adopted a particular EMR technology by time t.9 Thus identifies our main effect of interest. We have assumed that it is a normal i.i.d. variable and calculate heteroskedasticity-robust standard errors. We include two kinds of controls. First, Xit are controls for hospital characteristics that change over time such as inpatient days and outpatient visits (specified using a translog function), beds, type of hospital, etc. Second, Zi are controls for county-specific characteristics (such as population and income) that do not vary sufficiently over time for changes in their values to have much identifying power. However, the location-level characteristics do seem to have power to identify cost trends. Therefore, we interact these characteristics with the time trend to control for these trends. The full list of controls is provided in Table 2. Table 3 shows the results of this regression. For columns 1 to 7, the dependent variable is total administrative costs, as defined in the AHA data. For column 8 and 9, administrative costs are divided by the number of admittances to the hospital. Columns 1 to 3 use the specific EMR technologies that together we label “basic EMR”; columns 4 and 8 use the aggregated basic EMR measure (which whether the hospital has adopted any of the three technologies); column 5 and 6 use the EMR technologies that make up “advanced EMR”; columns 7 and 9 use the aggregated advanced EMR measure. The results suggest that, on average, EMR does not reduce costs. Instead, in most specifications, EMR is associated with a positive and significant increase in costs of about one to three percent. The increase appears to be slightly higher for advanced EMR than for basic EMR. Effects by time since adoption: As discussed above, a rich literature on IT productivity has documented that IT adoption affects productivity with a lag. Table 4 examines the extent to which the increase in costs is driven by initial adoption costs such as coinvention and learning new processes. Specifically, Table 4 splits the EMR variable into four pieces, based on time since adoption: (2)
Log(cit) =Xit+tZi+EMRit+EMRit-2 +EMRit-4 +EMRit-6 +t+i+it,
Because the data is biannual, the lags are t-2, t-4, and t-6. We therefore identify separate coefficients for the first period observed after adoption and for subsequent periods. These coefficients should be interpreted relative to the period before adoption.
9
As in Athey and Stern (2002), Hubbard (2003), and Forman, Goldfarb, and Greenstein (2012) we treat the diffusion of a new technology as an exogenous factor that leads to a change in economic outcomes, and then examine the plausibility of the exogeneity assumption.
13
In all cases, costs are significantly higher immediately after adoption, ranging from 1.1 percent higher for CDR to 3.1 percent higher for physician documentation and advanced EMR. After the first period, costs gradually return to the pre-adoption levels. Generally, the costs return to the pre-adoption levels faster for the basic than for the advanced technologies. In the case of CDR, costs appear to be lower four years after adoption. This is consistent with the literature on IT as a process innovation: initial adoption costs are high because of disruptions to established processes, over time these disruptions diminish, and more complicated technologies take more time to be effectively implemented. Table 5 sets up the specifications used in the remainder of the paper. Specifically, it focuses on the aggregate measure of basic and advanced EMR and it combines EMRit-2, EMRit-4, and EMRit-6 into one variable “adopt at least 2 years earlier”. As expected, the results are similar to Table 4. Effects by location: The literature on IT as a process innovation has emphasized that efficient use of IT based on the availability of complementary factors such as skilled labor, third-party software support and service, and infrastructure. To explore this hypothesis, we interact EMR adoption measures of the IT intensity of a location: (3)
Log(cit)=Xit+tZi+EMRit+EMRit-2+ +IT_INTENSEiEMRit +IT_INTENSEiEMRit-2+ +t+i+it,
where EMRit-2+ is a dummy variable for whether the hospital adopted EMR at least two years earlier and IT_INTENSEi is a measure of whether the location is IT-intensive. Specifically, Table 6 examines four distinct measures of IT-intensity: a dummy variable for whether the hospital is in a county is in the top quartile in terms of IT-using and IT-producing industry, a dummy variable for whether the hospital is in an MSA (found in prior work to be a rough but effective measure of local skills), the percentage of all firms in the county that had adopted basic IT by 2000, and the percentage of all firms in the county that had adopted advanced IT by 2000. Because the measures in columns 5 to 8 are continuous and the measures in the other column are discrete, a direct comparison of coefficient values is not appropriate. As a point of reference, the standard deviation for percentage that adopt basic IT is 0.3 and the standard deviation for advanced IT is 0.1. The first two rows show that costs generally increase in non-IT intensive counties, even for basic EMR and after two years. This increase is much lower in IT-intensive counties. In the first period after adoption, columns 1 and 9 show no significant rise in costs for basic EMR (using a Wald test of the sum of the coefficients in rows 1 and 3) and column 3 has marginal significance (p=0.09). For columns 5 and 7, significant disappears between one and two standard deviations. The differences increase after the initial adoption period: the sum of coefficients on basic EMR become significantly negative in IT-intensive locations. Interestingly, for advanced EMR, we see much less difference in the first period after adoption. Columns 1, 4, and 10 show no initial difference in costs between IT-intensive and other locations. Thus, the initial costs of adoption of advanced EMR, even when complementary inputs are available. However, two years after adoption, the impact of advanced EMR on costs is effectively zero in IT-intensive locations while it remains quite high in other locations. 14
Effects by hospital IT experience: Internal expertise can also mitigate the costs of adoption of a new process innovation. Table 7 examines the interaction in the following format: (4)
Log(cit)=Xit+tZi+EMRit+EMRit-2+ +HIT_EXPERIENCEiEMRit +HIT_EXPERIENCEiEMRit-2+ +t+i+it,
We use two measures of hospital IT experience: whether the hospital employed any programmers at the beginning of the sample (columns 1, 2, 5, and 6) and whether the hospital used at least one clinical software application at the beginning of our sample. Our results suggest a striking contrast to the effects of local expertise. Internal expertise appears to have little impact on the relationship between basic EMR and costs. Instead, it appears to have an substantial impact on reducing the cost increases from advanced EMR, particularly in the first period after adoption. Internal expertise therefore seems particularly important for the most advanced applications that might involve a great deal of coinvention to be successfully employed. Table 8 columns 1 and 2 show the results including the local IT intensity and hospital IT experience results in the same regression. Generally, the results of the separate analyses hold. For basic EMR adopters, (i) operating costs fall for hospitals in locations with high IT intensity, (ii) operating costs rise for other hospitals, and (iii) hospital IT experience has no significant relationship with costs. For advanced EMR adopters, (iv) operating costs do not rise significantly for hospitals in locations with high IT intensity, (v) operating costs rise for other hospitals, and (vi) the initial increase in operating costs after adoption is muted for hospitals with IT experience. The only qualitative change is that, for advanced EMR, the difference between IT intensive locations and other locations is not significant (though the magnitude of the coefficient is similar). Columns 3 and 4 of Table 8 show the three-way interaction between adoption, local IT intensity, and hospital IT experience. The many interactions introduce a great deal of noise to the estimation and little of significance appears. Mainly, we include these columns for completeness. Still, the signs in both columns, and the marginal significance in column 3 are somewhat suggestive that having IT experience is most effective in IT intensive locations (i.e. that they are complements). This contrasts with our prior work (Forman, Goldfarb, and Greenstein 2008) on the incentives to adopt frontier IT by businesses that suggests that urban locations can substitute for internal IT experience. Robustness, identification, and plausibility: Next, we explore the degree to which we can claim our main results are causal and general. There are three potential types of concerns. First, there might be an omitted variable correlated with EMR adoption and with costs. Second, it is possible that anticipated changes in costs drive EMR adoption (rather than EMR adoption driving changes in costs). Third, the large amount of missing data may mean that our sample is not representative. So far, to address these concerns, we have included hospital and time fixed effects as well as a very large set of covariates as controls. In order to address additional concerns we conduct three types of analyses, examining (i) the timing of the relationship between EMR adoption and cost changes, (ii) different types of costs, and (iii) robustness to alternative treatments of the missing variables. 15
In figure 2, we examine the timing of the relationship between EMR adoption and changes in costs. Specifically, we run the equation (2) above, but add variables for 2 years before adoption, 4 years before adoption, and 8 years before adoption. Figure 2a shows that costs seem to peak in the adoption period, and then subsequently fall. More interestingly, figures 2b and 2c demonstrate distinct effects for IT intensive and non IT intensive locations, defined by the top quartile of counties in terms of IT-intensive industry. Figure 2b examines basic EMR adoption and figure 2c examines advanced EMR adoption. Prior to adoption, the costs follow similar patterns. During and after the initial adoption, however, the costs in non-IT intensive locations rise while the costs in IT-intensive locations fall substantially. The full results for these regressions are shown in appendix table 1. Figure 2 therefore suggests that the timing of the impact of EMR is appropriate: there does not appear to be a noticeable omitted variable driving In table 9, we compare diagnostic radiology costs and nursing costs. Many observers have argued the CDR, CDS, and CPOE should be particularly effective at reducing diagnostic radiology costs (e.g. Wager, Wickham Lee, and Glaser 2009) The first five columns document that, even without the interactions with local or internal expertise, such costs experience no initial increase and then appear to fall. In contrast, some costs might be expected to rise as a consequence of particular EMR technologies, most notably nursing costs through increase in administration and clerical work (Sidorov 2006). Columns 6 through 10 of Table 9 confirm this hypothesis: nursing costs generally rise, particularly a few years after adoption when the EMRs generally reduce other costs. Broadly, we view these results as consistent with the impact of EMR being driven by the places where we would expect to see an impact, rather than a spurious relationship to all costs. Third, address the concern that several of our key covariates and controls are missing in some years for some hospitals. In particular, we lack data on EMR adoption for 14 to 35 percent of the sample, depending on the specification. In addition, we lack data on discharges and hospital types for a similar fraction of the data. In the main specification, we drop observations that are missing, implicitly assuming that they are missing-at-random. We conduct two types of analysis in order to determine whether the missing data generates a bias. First, for the missing controls (on discharges and hospital types), we create a new dummy variable for when these observations are missing. While Allison (2002) points out that this method can lead to biased coefficients if the missing data covariate does not represent a real characteristic of the hospitals, even then the signs do not change in linear models. Second, to address the concerns on the missing EMR data, we show that the results hold with two opposite assumptions: if we set all missing values to zero or if we set all missing values to one. This is suggestive that the missing data does not create a systematic bias. Results are reported in the appendix. VI. Conclusion Drawing on a variety of data sources on IT, EMR, local demographics, and hospital characteristics, we have shown evidence of complementarities between EMR adoption and other inputs. Specifically, while on average, EMR adoption seems be to associated with an increase in costs, this hides important heterogeneity over time, across technologies, across locations, and across hospitals. We find that both basic and especially advanced EMR adoption are initially 16
associated with a rise in costs but this initial increase in costs is mitigated in hospitals with some internal information technology expertise. After two years, hospitals in locations that have local information technology expertise experience a decrease in costs after adopting basic EMR, and no increase or decrease in costs after adopting advanced EMR. In contrast, hospitals in other locations experience a sharp increase in costs, even after several years. As with any empirical work, our analysis has a number of limitations. First, we observe only a subset of the medical practices in the United States. Smaller doctors’ offices, outpatient clinics, nursing homes, and other medical practices may have had a different experience. While we believe it is likely that the general principles of business process innovation would apply broadly, our evidence is specific to hospitals. Second, we do not directly address the question of why hospitals adopt if their costs do not fall. It might be due to misconceptions, expected benefits that we do not measure, or something else. We have tried to address the endogeneity of this adoption through various techniques, but we cannot completely rule out the possibility that adopting hospitals in IT-intensive locations adopt because they expect their costs to fall and therefore they have resources. Third, and relatedly, it is possible that hospitals outside ITintensive locations experience a sharp increase in benefits such as clinical outcomes and reduced errors (though the evidence in the literature is mixed on whether such benefits accrue). Thus our findings on reduced costs only tell part of the story. Despite these limitations, we believe our results linking the mixed evidence on the success of EMR in reducing costs with the literature on IT as a business process innovation help inform the discussion on the “trillion dollar conundrum”: the (perhaps missing) link between healthcare IT and healthcare costs.
17
References Agha, Leila (2012), “The Effects of Health Information Technology on the Costs and Quality of Medical Care,” Working paper. https://sites.google.com/site/leilaagha/research (accessed January, 2012). Allison, Paul. 2002. Missing Data, Sage Publications: Thousand Oaks, CA. Arora, Ashish and Chris Forman (2007) “Proximity and Information Technology Outsourcing: How Local are IT Services Markets” Journal of Management Information System 24(2): 73-102. Athey, S., and S. Stern, 2002, The Impact of Information Technology on Emergency Health Care Outcomes, RAND Journal of Economics 33, 399-432. Attewell, P., 1992, “Technology Diffusion and Organizational Learning: The Case of Business Computing,” Organizational Science, 3, 1–19. Autor, David, Frank Levy, and Richard J. Murnane. 2003. The Skill Content of Recent Technological Change: An Empirical Exploration. Quarterly Journal of Economics 118(4), 1279–334. Banker, R. and S. Slaughter. 1997. “A Field Study of Scale Economies in Software Maintenance,” Management Science 43: 1709–1725. Barnes, B.H. and T.B. Bollinger. 1991. “Making reuse cost-effective.” IEEE Software 8(1): 13– 24. Beaudry, Paul, Mark Doms, and Ethan Lewis. 2006. Endogenous Skill Bias in Technology Adoption: City-Level Evidence from the IT Revolution. Federal Reserve Bank of San Francisco. Working Paper #06-24 Black, S. E., and L. M. Lynch, 2001, How To Compete: The Impact of Workplace Practices and Information Technology on Productivity, Review of Economics and Statistics 83, 434445. Bloom, N., L. Garicano, R. Sadun, and J. Van Reenen. 2009. The distinct effects of Information Technology and Communication Technology on Firm Organization. NBER Working Paper #14975. Bloom, Nicholas, Raffaella Sadun, and John Van Reenen. 2012. Americans Do IT Better: US Multinationals and the Productivity Miracle. American Economic Review 102(1). Blundell, R. and Bond, S. 2000. GMM estimation with Persistent Panel Data: An Application to Production Functions. Econometric Review 19(3): 321‐40. Borzokowski, R. (2009), “Measuring the Cost Impact of Hospital Information Systems: 19871994” Journal of Health Economics 28: 938-49. 18
Bresnahan, T. F., E. Brynjolfsson, and L. M. Hitt. 2002. Information Technology, Workplace Organization, and the Demand for Skilled Labor: Firm-Level Evidence. Quarterly Journal of Economics 117(1): 339-376. Bresnahan, T. and S. Greenstein. 1996. Technical Progress and Co-invention in Computing and in the Uses of Computers. Brookings Papers on Economic Activity, Microeconomics 1996; 1-83. Brynjolfsson, E. and L. Hitt. 2003. Computing Productivity: Firm-Level Evidence. The Review of Economics and Statistics 85(4): 793-808. Buntin, M. and Cutler, D. (2009) “The Two Trillion Dollar Solution” Center for American Progress. http://www.americanprogress.org/issues/2009/06/pdf/2trillion_solution.pdf Centers for Medicare & Medicaid Services. 2010. National Health Expenditure Projections 20102020. Available at https://www.cms.gov/NationalHealthExpendData/downloads/proj2010.pdf . Accessed on February 14, 2012. Cohen, W. and D. Levinthal, 1990, “Absorptive Capacity: A New Perspective on Learning and Innovation,” Administrative Science Quarterly, 35, 128–152. Congressional Budget Office. 2008. Evidences on the Costs and Benefits of Healthcare Information Technology. Available at http://www.cbo.gov/ftpdocs/91xx/doc9168/maintext.3.1.shtml. Accessed on February 14, 2012. Fonkych, K. and R. Taylor, 2005, The State and Pattern of Health Information Technology Adoption Santa Monica, RAND Corporation Forman, Chris and Avi Goldfarb. 2006. Diffusion of Information and Communication Technologies to Businesses. In Handbook of Information Systems, Volume 1: Economics and Information Systems, ed. Terrence Hendershott. 1–52. Amsterdam: Elsevier. Forman, Chris, Avi Goldfarb, and Shane Greenstein. 2005. How Did Location Affect the Adoption of the Commercial Internet? Global Village vs Urban Density. Journal of Urban Economics, 58(3): 389–420. Forman, Chris, Avi Goldfarb, and Shane Greenstein (2008), “Understanding Inputs into Innovation: Do Cities Substitute for Internal Firm Resources?” Journal of Economics and Management Strategy. 295-316. Forman, Chris, Avi Goldfarb, and Shane Greenstein. 2012. The Internet and Local Wages: A Puzzle. American Economic Review, 102(1) 556-575. Furukawa, M., Raghu, T. and B. Shao, 2010, “Electronic Medical Records, Nurse Staffing, and Nurse-Sensitive Patient Outcomes: Evidence from California Hospitals, 1998-2007. Health Services Research 45(4): 941-962.
19
Greenstein, Shane, and Ryan McDevitt (2011), “The Broadband Bonus: Estimating Broadband Internet’s Economic Value,” Telecommunications Policy, 35, pp 617-632. Hillestad R, Bigelow J, Bower A, Girosi F, Meili R, Scoville R, and Taylor R, 2005, “Can Electronic Medical Record Systems Transform Healthcare? An Assessment of Potential Health Benefits, Savings, and Costs,” Health Affairs, 24(5): 1103-17. HIMSS Analytics. 2011. Essentials of the U.S. Hospital IT Market, 6th Edition. Hubbard, T. N., 2003, Information, Decisions, and Productivity: On-Board Computers and Capacity Utilization in Trucking, American Economic Review 93, 1328-1353. Jorgenson, Dale W., Mun S. Ho, and Kevin J. Stiroh. 2005. Productivity, Volume 3: Information Technology and the American Growth Resurgence. Cambridge, MA: MIT Press. Kazley, A. and Y. Ozcan, 2007, “Organizational and Environmental Determinants of Hospital EMR Adoption: A National Study” Journal of Medical Systems 31: 375-84. Kolko, Jed. 2002. Silicon Mountains, Silicon Molehills: Geographic concentration and convergence of internet industries in the US. Information Economics and Policy 14(2): 211–32. Manski, C., 1993, “Identification of Endogenous Social Effects: The Reflection Problem” Review of Economic Studies, 60(3): 531-42 McCullough, J., 2008, “The Adoption of Hospital Information Systems” Health Economics, 17: 649-64. Miller, A. and C. Tucker, 2009, “Privacy Protection and Technology Diffusion: The Case of Electronic Medical Records” Management Science 55(7): 1077-93. Miller, A. and C. Tucker. 2012. “Can Healthcare Information Technology Save Babies?”, Journal of Political Economy, Vol. 119 No. 2, April 2011, pp. 289-324 Parente, S. and L. Van Horn, 2006, “Valuing Hospital Investment in Information Technology: Does Governance Make a Difference?” Health Care Financing Review 28(2): 31-43 Sidorov, J., 2006, “It Ain’t Necessarily So: The Electronic Health Record and the Unlikely Prospect of Reducing Health Care Costs.” Health Affairs 25(4): 1079-85. U.S. Department of Health and Human Services, 2011, “Surveys Show Significant Proportions of Hospitals and Doctors Already Plan to Adopt Electronic Health Records and Qualify for Federal Incentive Payments” HHS Press Release, 1/13/ 2011. Wager, Karen A., Frances Wickham Lee, and John P. Glaser. 2009. Health Care Information Systems. Second Edition. Wiley/Jossey-Bass, San Francisco.
20
Table 1: Types of EMR and Hospital Adoption Rates EMR
Description
Clinical Data Repository
Real time database that consolidates clinical data to create a unified patient medical record Clinical Decision Support Uses patient data to generate diagnostic and/or treatment advice Order Entry Provides electronic forms to streamline hospital operations (replacing faxes and paper forms Computerized Physician Electronic entry of physician treatment orders that can be Order Entry communicated to the pharmacy, lab, and other departments Physician Documentation Allows physicians to transition from written to electronic notes
% of Hospitals Adopting 1996 2008 .134 .775 .109
.716
.166
.823
.009
.222
.026
.211
Table 2a: Summary statistics for 2008 Variable
Mean
Std. Dev.
Min
EMR MEASURES CDR CDSS Order entry Basic EMR adoption (CDR, CDSS, or order entry) CPOE Physician documentation Advanced EMR adoption (CPOE or Physician doc’n) COST MEASURES Log total costs Log total costs per admit Log lab diagnostic costs Log nursing costs HOSPITAL‐LEVEL CONTROLS Log inpatient days Log outpatient visits Log total hospital beds Independent practice association hospital Management service organization hospital Equity model hospital Foundation hospital Log admissions Births (000s) Full time physicians and dentists (000s) Percent births For‐profit ownership Non‐secular nonprofit ownership Non‐profit church ownership Number of discharges Medicare (000s) Number of discharges Medicaid (000s) Number of discharges total (000s) Residency or Member of Council Teaching Hospitals Vertically integrated with doctors Vert. integ. with doctors (excl. integrated salary model) FT physicians / total hospital beds LOCATION‐LEVEL CONTROLS MSA dummy Log population in 2000 census % Hispanic in 2000 census % Black in 2000 census % age 65+ in 2000 census % age 25‐64 in 2000 census % university education in 2000 census Log median home value in 2000 census Log median household income in 2000 census OTHER VARIABLES USED Top quartile county IT intensive industry % local businesses adopted basic internet by 2000 % local businesses adopted advanced internet by 2000 Had at least one programmer in 1996 Had at least one clinical software application in 1996
0.775 0.716 0.823 0.835 0.222 0.211 0.283
0.418 0.451 0.382 0.372 0.415 0.408 0.451
17.765 9.372 15.203 13.225
1.347 0.372 1.158 1.362
0 0 0 0 0 0 0 14.366 7.008 9.821 4.820
9.834 11.118 4.598 0.111 0.062 0.014 0.031 8.110 0.893 0.019 0.101 0.142 0.489 0.109 2.324 0.925 6.608 0.171 0.602 0.357 0.094
1.424 1.265 1.090 0.314 0.241 0.119 0.173 1.463 1.371 0.088 0.104 0.349 0.500 0.312 2.825 1.646 8.773 0.377 0.490 0.479 0.174
2.398 0.693 0.693 0 0 0 0 1.099 0 0 0 0 0 0 1 1 1 0 0 0 0
0.527 11.566 0.091 0.101 0.139 0.851 0.134 148.431 136.982
0.499 1.848 0.136 0.138 0.040 0.047 0.059 6.137 3.146
0.428 0.804 0.104 0.195 0.670
0.495 0.315 0.101 0.396 0.470
Max
# obs.
1 1 1 1 1 1 1 21.873 13.812 18.650 17.243
2982 2704 3163 2255 3648 3593 3305 3510 2295 2938 3438
13.434 15.022 7.698 1 1 1 1 11.619 17.203 2.170 2.241 1 1 1 22.063 29.479 70.133 1 1 1 2.531
4247 4232 4247 3466 3469 3467 3467 4247 4247 4247 4247 4323 4323 4323 2329 2329 2329 4323 4323 4323 4247
0 1 7.336 16.069 0 0.981 0 0.861 0.028 0.347 0.455 1.051 0.037 0.402 131.170 170.513 126.063 147.235 0 1 0 1 0 1 0 1 0 1
4283 4283 4283 4283 4283 4283 4283 4283 4283 4107 4107 4107 1563 1556
Table 2b: Summary statistics on changes from 1996 to 2008 Variable Log total costs CDR CDSS Order entry Basic EMR adoption (CDR, CDSS, or order entry) CPOE Physician documentation Advanced EMR adoption (CPOE or Physician doc’n)
Mean Std. Dev. 0.791 0.386 0.661 0.474 0.607 0.488 0.657 0.475 0.680 0.467 0.213 0.410 0.185 0.388 0.255 0.436
Min Max # obs. ‐1.316 3.016 2218 0 1 2976 0 1 2700 0 1 3156 0 1 2251 0 1 3643 0 1 3587 0 1 3300
Table 3: Main effects by technology
(1) Log total costs
(2) Log total costs
(3) Log total costs
(4) Log total costs
(5) Log total costs
(6) Log total costs
Technology
CDR
CDSS
Order entry
Basic EMR adoption
CPOE
Adopted EMR
0.0080 (0.0057) 12070 2885 0.80 ‐0.3178 (0.0727)*** ‐0.0745 (0.0519) 0.0188 (0.0045)*** 0.0061 (0.0015)*** ‐0.0006 (0.0044) 0.0868 (0.0158)*** 0.0090 (0.0073) 0.0161 (0.0076)** ‐0.0046 (0.0179) ‐0.0449 (0.0145)*** 0.0099 (0.0254)
0.0181 (0.0061)*** 10828 2596 0.80 ‐0.2661 (0.0880)*** ‐0.0857 (0.0599) 0.0156 (0.0047)*** 0.0060 (0.0014)*** 0.0009 (0.0049) 0.0918 (0.0163)*** 0.0023 (0.0071) 0.0176 (0.0073)** ‐0.0178 (0.0198) ‐0.0370 (0.0135)*** 0.0078 (0.0264)
0.0095 (0.0055)*
0.0127 (0.0066)*
0.0126 (0.0076)*
12877 3057 0.80
9122 2178 0.80
14670 3515 0.79
‐0.3499 (0.1032)*** ‐0.1654 (0.0738)** 0.0187 (0.0051)*** 0.0090 (0.0036)** 0.0016 (0.0066) 0.0908 (0.0176)*** 0.0024 (0.0072) 0.0086 (0.0075) ‐0.0018 (0.0177) ‐0.0328 (0.0143)** 0.0201 (0.0259)
‐0.2939 (0.0895)*** ‐0.1389 (0.0629)** 0.0197 (0.0055)*** 0.0104 (0.0040)*** ‐0.0036 (0.0067) 0.0851 (0.0179)*** 0.0016 (0.0078) 0.0187 (0.0080)** ‐0.0046 (0.0182) ‐0.0333 (0.0151)** 0.0124 (0.0263)
‐0.3807 (0.0779)*** ‐0.1177 (0.0622)* 0.0198 (0.0039)*** 0.0059 (0.0015)*** 0.0037 (0.0050) 0.0901 (0.0159)*** 0.0031 (0.0067) 0.0098 (0.0071) ‐0.0132 (0.0176) ‐0.0391 (0.0143)*** ‐0.0002 (0.0236)
Physician Advanced documentation EMR adoption 0.0295 0.0230 (0.0083)*** (0.0076)*** 14446 13193 3462 3175 0.79 0.79 ‐0.3851 ‐0.3953 (0.0782)*** (0.0794)*** ‐0.1577 ‐0.1599 (0.0609)*** (0.0621)** 0.0187 0.0193 (0.0040)*** (0.0041)*** 0.0072 0.0071 (0.0016)*** (0.0015)*** 0.0055 0.0060 (0.0050) (0.0051) 0.0879 0.0903 (0.0159)*** (0.0166)*** 0.0065 0.0047 (0.0067) (0.0071) 0.0088 0.0091 (0.0072) (0.0076) ‐0.0047 ‐0.0107 (0.0169) (0.0194) ‐0.0354 ‐0.0316 (0.0122)*** (0.0140)** ‐0.0052 ‐0.0006 (0.0236) (0.0244)
Observations # of hospitals R‐squared CONTROLS Log inpatient days Log outpatient visits Log inpatient days x Log inpatient days Log outpatient visits x Log outpatient visits Log inpatient days x Log outpatient visits Log total hospital beds Independent practice association hospital Management service organization hospital Equity model hospital Foundation hospital Log admissions
(7) Log total costs
(8) Log total costs per admit Basic EMR adoption 0.0115 (0.0068)*
(9) Log total costs per admit Advanced EMR adoption 0.0258 (0.0078)***
7687 1564 0.78
11082 2268 0.77
‐0.4715 (0.1337)*** ‐0.0814 (0.0894) 0.0293 (0.0081)*** 0.0099 (0.0049)** ‐0.0079 (0.0090) 0.0978 (0.0192)*** 0.0037 (0.0080) 0.0166 (0.0079)** ‐0.0083 (0.0180) ‐0.0366 (0.0160)** ‐0.9432 (0.0374)***
‐0.4493 (0.0921)*** ‐0.0820 (0.0898) 0.0239 (0.0047)*** 0.0065 (0.0015)*** ‐0.0001 (0.0072) 0.1046 (0.0171)*** 0.0093 (0.0072) 0.0082 (0.0076) ‐0.0148 (0.0192) ‐0.0336 (0.0150)** ‐0.9642 (0.0327)***
Births (000s)
0.0050 (0.0074) Full time physicians and ‐0.2123 dentists (000s) (0.1214)* Percent births 0.1413 (0.1092) For‐profit ownership ‐0.0282 (0.0316) Non‐secular nonprofit 0.0314 ownership (0.0229) Non‐profit church 0.0730 ownership (0.0321)** Log number of 0.1511 discharges Medicare (0.0413)*** 0.0050 Log number of discharges Medicaid (0.0023)** Log number of 0.2141 discharges total (0.0445)*** Residency or Member of ‐0.0023 Council Teaching Hosps (0.0118) 0.0038 Vertically integrated (0.0073) with doctors Vert. integ. w drs (excl. 0.0063 integrated salary mdl) (0.0092) FT physicians / total 0.1915 hospital beds (0.0595)*** Year 1998 ‐0.0488 (0.0740) Year 2000 ‐0.1071 (0.1478) Year 2002 ‐0.1298 (0.2211) Year 2004 ‐0.1791 (0.2949) Year 2006 ‐0.1967 (0.3689) Year 2008 ‐0.2267
0.0028 (0.0086) ‐0.3271 (0.1401)** 0.2361 (0.1308)* ‐0.0367 (0.0363) 0.0172 (0.0250) 0.0586 (0.0345)* 0.1506 (0.0445)*** 0.0077 (0.0026)*** 0.2421 (0.0498)*** ‐0.0007 (0.0111) 0.0110 (0.0074) 0.0023 (0.0094) 0.2458 (0.0700)*** ‐0.0540 (0.0747) ‐0.1150 (0.1489) ‐0.1417 (0.2229) ‐0.1917 (0.2974) ‐0.2151 (0.3720) ‐0.2497
0.0049 (0.0078) ‐0.2402 (0.1171)** 0.1752 (0.1183) ‐0.0136 (0.0284) 0.0347 (0.0213) 0.0909 (0.0293)*** 0.1429 (0.0350)*** 0.0057 (0.0023)** 0.2133 (0.0396)*** ‐0.0033 (0.0102) 0.0057 (0.0072) 0.0034 (0.0090) 0.2018 (0.0591)*** ‐0.0625 (0.0703) ‐0.1333 (0.1404) ‐0.1691 (0.2101) ‐0.2329 (0.2803) ‐0.2664 (0.3506) ‐0.3104
0.0033 (0.0084) ‐0.3757 (0.1860)** 0.1159 (0.1155) ‐0.0216 (0.0371) 0.0177 (0.0276) 0.0701 (0.0372)* 0.1374 (0.0444)*** 0.0076 (0.0025)*** 0.2331 (0.0519)*** 0.0072 (0.0119) 0.0111 (0.0082) ‐0.0018 (0.0105) 0.2470 (0.0817)*** ‐0.0288 (0.0803) ‐0.0630 (0.1598) ‐0.0650 (0.2393) ‐0.0926 (0.3193) ‐0.0924 (0.3994) ‐0.1010
0.0065 (0.0071) ‐0.3492 (0.1134)*** 0.1549 (0.0950) 0.0026 (0.0268) 0.0392 (0.0206)* 0.0825 (0.0293)*** 0.1269 (0.0296)*** 0.0054 (0.0022)** 0.2235 (0.0341)*** 0.0015 (0.0105) 0.0056 (0.0066) 0.0058 (0.0082) 0.2614 (0.0650)*** ‐0.0770 (0.0668) ‐0.1590 (0.1335) ‐0.2071 (0.1999) ‐0.2798 (0.2666) ‐0.3267 (0.3334) ‐0.3853
0.0064 (0.0072) ‐0.2632 (0.1121)** 0.1762 (0.0926)* ‐0.0071 (0.0277) 0.0325 (0.0207) 0.0793 (0.0290)*** 0.1371 (0.0332)*** 0.0040 (0.0022)* 0.2243 (0.0367)*** 0.0030 (0.0100) 0.0045 (0.0066) 0.0108 (0.0082) 0.2188 (0.0568)*** ‐0.0886 (0.0677) ‐0.1833 (0.1355) ‐0.2425 (0.2028) ‐0.3308 (0.2705) ‐0.3875 (0.3382) ‐0.4566
0.0061 (0.0074) ‐0.3627 (0.1440)** 0.1752 (0.0968)* ‐0.0046 (0.0287) 0.0345 (0.0218) 0.0829 (0.0301)*** 0.1333 (0.0338)*** 0.0044 (0.0022)** 0.2196 (0.0376)*** 0.0067 (0.0104) 0.0053 (0.0069) 0.0102 (0.0086) 0.2595 (0.0713)*** ‐0.1035 (0.0712) ‐0.2116 (0.1424) ‐0.2853 (0.2132) ‐0.3848 (0.2844) ‐0.4568 (0.3556) ‐0.5419
‐0.0032 (0.0093) ‐0.3180 (0.2232) 0.1793 (0.1327) ‐0.0065 (0.0366) 0.0322 (0.0281) 0.0834 (0.0380)** 0.1356 (0.0468)*** 0.0071 (0.0025)*** 0.2384 (0.0581)*** 0.0076 (0.0120) 0.0065 (0.0086) 0.0017 (0.0110) 0.2177 (0.1015)** ‐0.0146 (0.0804) ‐0.0357 (0.1599) ‐0.0241 (0.2394) ‐0.0280 (0.3194) ‐0.0243 (0.3996) ‐0.0365
0.0012 (0.0080) ‐0.3567 (0.1647)** 0.1954 (0.1118)* ‐0.0080 (0.0299) 0.0219 (0.0224) 0.0750 (0.0310)** 0.1299 (0.0350)*** 0.0037 (0.0023) 0.2437 (0.0428)*** 0.0052 (0.0102) 0.0008 (0.0074) 0.0122 (0.0090) 0.2542 (0.0832)*** ‐0.0837 (0.0715) ‐0.1736 (0.1430) ‐0.2281 (0.2141) ‐0.2987 (0.2855) ‐0.3618 (0.3571) ‐0.4452
(0.4436) (0.4475) (0.4213) (0.4804) (0.4006) (0.4063) (0.4272) (0.4810) (0.4290) ‐0.0063 ‐0.0049 ‐0.0056 ‐0.0051 ‐0.0051 ‐0.0057 ‐0.0059 ‐0.0046 ‐0.0057 (0.0018)*** (0.0018)*** (0.0017)*** (0.0019)*** (0.0017)*** (0.0017)*** (0.0018)*** (0.0019)** (0.0018)*** Log population in 2000 ‐0.0008 ‐0.0020 ‐0.0011 ‐0.0017 ‐0.0015 ‐0.0017 ‐0.0018 ‐0.0013 ‐0.0014 census x year (0.0007) (0.0007)*** (0.0007)* (0.0007)** (0.0006)** (0.0006)*** (0.0007)*** (0.0007)* (0.0007)** ‐0.0061 ‐0.0064 ‐0.0058 ‐0.0093 ‐0.0066 % Hispanic in 2000 ‐0.0095 ‐0.0086 ‐0.0091 ‐0.0091 (0.0054)* (0.0054) (0.0052)* (0.0057) (0.0050) (0.0050) (0.0053) (0.0058) (0.0053) census x year % Black in 2000 census ‐0.0149 ‐0.0143 ‐0.0155 ‐0.0149 ‐0.0126 ‐0.0122 ‐0.0101 ‐0.0154 ‐0.0109 x year (0.0046)*** (0.0045)*** (0.0045)*** (0.0049)*** (0.0043)*** (0.0044)*** (0.0045)** (0.0049)*** (0.0045)** % age 65+ in 2000 ‐0.0470 ‐0.0310 ‐0.0400 ‐0.0388 ‐0.0382 ‐0.0407 ‐0.0373 ‐0.0462 ‐0.0448 (0.0194)** (0.0221)* (0.0187)** (0.0191)** (0.0198)* (0.0229)** (0.0203)** census x year (0.0200)** (0.0207) % age 25‐64 in 2000 ‐0.0294 ‐0.0359 ‐0.0280 ‐0.0383 ‐0.0316 ‐0.0151 ‐0.0157 ‐0.0361 ‐0.0135 census x year (0.0157)* (0.0154)** (0.0154)* (0.0158)** (0.0140)** (0.0145) (0.0150) (0.0161)** (0.0152) % university education ‐0.0088 ‐0.0107 ‐0.0116 ‐0.0066 ‐0.0088 ‐0.0063 ‐0.0045 ‐0.0067 ‐0.0051 in 2000 census x year (0.0150) (0.0154) (0.0139) (0.0166) (0.0133) (0.0137) (0.0144) (0.0168) (0.0146) 0.0121 0.0108 0.0102 0.0123 0.0124 0.0130 0.0093 0.0124 Log median home value 0.0098 in 2000 census x year (0.0028)*** (0.0027)*** (0.0026)*** (0.0030)*** (0.0025)*** (0.0025)*** (0.0026)*** (0.0030)*** (0.0026)*** Log median hh income in 0.0017 0.0009 0.0014 0.0017 0.0009 0.0003 0.0004 0.0014 ‐0.0006 2000 census x year (0.0048) (0.0046) (0.0045) (0.0050) (0.0044) (0.0043) (0.0046) (0.0051) (0.0047) Constant 15.2452 14.7712 15.9365 15.3679 15.9130 16.1138 16.1379 15.4462 15.4295 (0.4685)*** (0.6793)*** (0.8237)*** (0.5676)*** (0.6475)*** (0.6382)*** (0.6438)*** (0.7754)*** (0.8342)*** Unit of observation is a hospital‐year. Data biannually from 1996 to 2008. Regressions include hospital‐specific fixed effects, differenced out at means. MSA dummy x year
Table 4: Main effects by technology, by years since adoption
(1) Log total costs
(2) Log total costs
(3) Log total costs
Technology
CDR
CDSS
Order entry Basic EMR adoption
CPOE
0.0169 (0.0063)*** 0.0101 (0.0077) 0.0049 (0.0098) ‐0.0052 (0.0124)
0.0101 (0.0059)* ‐0.0021 (0.0073) ‐0.0123 (0.0091) ‐0.0135 (0.0120)
0.0207 (0.0056)*** 0.0065 (0.0060) 0.0012 (0.0059) 0.0040 (0.0068)
10828 2596 0.80
12877 3057 0.80
‐0.2667 (0.0880)*** ‐0.0846 (0.0597) 0.0156 (0.0047)*** 0.0060 (0.0014)*** 0.0009 (0.0049) 0.0923 (0.0163)*** 0.0020 (0.0071) 0.0171 (0.0073)**
‐0.3505 (0.1031)*** ‐0.1674 (0.0734)** 0.0187 (0.0051)*** 0.0091 (0.0036)** 0.0016 (0.0065) 0.0906 (0.0176)*** 0.0016 (0.0072) 0.0080 (0.0075)
Adopt in previous 2 year period Adopt 2 to 4 years earlier
0.0108 (0.0059)* ‐0.0078 (0.0073) Adopt 4 to 6 years earlier ‐0.0168 (0.0091)* Adopt at least 6 years ‐0.0218 (0.0123)* earlier Observations 12070 # of hospitals 2885 R‐squared 0.80 CONTROLS Log inpatient days ‐0.3160 (0.0722)*** Log outpatient visits ‐0.0745 (0.0516) Log inpatient days x Log 0.0187 inpatient days (0.0045)*** Log outpatient visits x Log 0.0062 outpatient visits (0.0015)*** Log inpatient days x Log ‐0.0006 outpatient visits (0.0044) Log total hospital beds 0.0876 (0.0158)*** Independent practice 0.0085 association hospital (0.0072) Management service 0.0154 organization hospital (0.0076)**
(4) Log total costs
(5) Log total costs
(6) Log total costs
(7) Log total costs
(8) Log total costs per admit Physician Advanced EMR Basic EMR documentation adoption adoption
0.0154 (0.0077)** 0.0073 (0.0097) 0.0112 (0.0123) ‐0.0139 (0.0189)
0.0313 (0.0082)*** 0.0325 (0.0103)*** 0.0177 (0.0128) ‐0.0105 (0.0179)
0.0312 (0.0073)*** 0.0196 (0.0093)** 0.0180 (0.0109)* ‐0.0112 (0.0143)
0.0194 (0.0058)*** 0.0030 (0.0061) 0.0003 (0.0059) 0.0022 (0.0069)
(9) Log total costs per admit Advanced EMR adoption 0.0329 (0.0075)*** 0.0201 (0.0094)** 0.0213 (0.0110)* ‐0.0056 (0.0141)
9122 2178 0.80
14670 3515 0.79
14446 3462 0.79
13193 3175 0.79
7687 1564 0.78
11082 2268 0.77
‐0.2959 (0.0897)*** ‐0.1380 (0.0623)** 0.0198 (0.0055)*** 0.0104 (0.0040)*** ‐0.0037 (0.0067) 0.0854 (0.0179)*** 0.0012 (0.0078) 0.0182 (0.0080)**
‐0.3807 (0.0779)*** ‐0.1190 (0.0622)* 0.0198 (0.0039)*** 0.0059 (0.0015)*** 0.0038 (0.0050) 0.0901 (0.0159)*** 0.0030 (0.0067) 0.0100 (0.0071)
‐0.3851 (0.0782)*** ‐0.1579 (0.0605)*** 0.0186 (0.0040)*** 0.0072 (0.0015)*** 0.0057 (0.0050) 0.0887 (0.0159)*** 0.0063 (0.0067) 0.0087 (0.0072)
‐0.3959 (0.0793)*** ‐0.1607 (0.0618)*** 0.0192 (0.0041)*** 0.0071 (0.0015)*** 0.0062 (0.0051) 0.0908 (0.0166)*** 0.0045 (0.0071) 0.0090 (0.0076)
‐0.4765 (0.1342)*** ‐0.0803 (0.0888) 0.0297 (0.0081)*** 0.0099 (0.0048)** ‐0.0081 (0.0090) 0.0978 (0.0193)*** 0.0032 (0.0079) 0.0162 (0.0079)**
‐0.4502 (0.0920)*** ‐0.0834 (0.0894) 0.0238 (0.0047)*** 0.0065 (0.0015)*** 0.0002 (0.0072) 0.1049 (0.0171)*** 0.0091 (0.0072) 0.0081 (0.0076)
Equity model hospital Foundation hospital Log admissions Births (000s) Full time physicians and dentists (000s) Percent births For‐profit ownership Non‐secular nonprofit ownership Non‐profit church ownership Log number of discharges Medicare Log number of discharges Medicaid Log number of discharges total Residency or Member of Council Teaching Hosps Vertically integrated with doctors Vert. integ. w drs (excl. integrated salary mdl) FT physicians / total hospital beds Year 1998 Year 2000 Year 2002
‐0.0046 (0.0180) ‐0.0465 (0.0145)*** 0.0107 (0.0254) 0.0050 (0.0074) ‐0.2093 (0.1210)* 0.1412 (0.1095) ‐0.0306 (0.0315) 0.0306 (0.0227) 0.0718 (0.0321)** 0.1507 (0.0413)*** 0.0050 (0.0023)** 0.2155 (0.0445)*** ‐0.0012 (0.0118) 0.0038 (0.0073) 0.0058 (0.0092) 0.1904 (0.0592)*** ‐0.0447 (0.0739) ‐0.0975 (0.1475) ‐0.1152
‐0.0177 (0.0199) ‐0.0372 (0.0135)*** 0.0082 (0.0265) 0.0031 (0.0086) ‐0.3238 (0.1399)** 0.2371 (0.1312)* ‐0.0365 (0.0362) 0.0179 (0.0249) 0.0582 (0.0344)* 0.1503 (0.0444)*** 0.0077 (0.0026)*** 0.2427 (0.0497)*** 0.0001 (0.0112) 0.0104 (0.0073) 0.0023 (0.0094) 0.2443 (0.0699)*** ‐0.0487 (0.0745) ‐0.1036 (0.1485) ‐0.1233
‐0.0019 (0.0178) ‐0.0335 (0.0143)** 0.0205 (0.0259) 0.0048 (0.0078) ‐0.2419 (0.1170)** 0.1789 (0.1183) ‐0.0146 (0.0284) 0.0347 (0.0212) 0.0905 (0.0294)*** 0.1429 (0.0351)*** 0.0056 (0.0023)** 0.2135 (0.0396)*** ‐0.0027 (0.0102) 0.0058 (0.0072) 0.0032 (0.0090) 0.2022 (0.0590)*** ‐0.0579 (0.0703) ‐0.1225 (0.1403) ‐0.1533
‐0.0046 (0.0184) ‐0.0337 (0.0151)** 0.0129 (0.0264) 0.0033 (0.0084) ‐0.3792 (0.1856)** 0.1177 (0.1157) ‐0.0215 (0.0371) 0.0185 (0.0275) 0.0701 (0.0371)* 0.1373 (0.0444)*** 0.0075 (0.0025)*** 0.2332 (0.0519)*** 0.0082 (0.0119) 0.0108 (0.0082) ‐0.0018 (0.0105) 0.2479 (0.0817)*** ‐0.0230 (0.0800) ‐0.0502 (0.1593) ‐0.0473
‐0.0131 (0.0176) ‐0.0394 (0.0143)*** ‐0.0001 (0.0236) 0.0067 (0.0071) ‐0.3481 (0.1133)*** 0.1545 (0.0949) 0.0022 (0.0268) 0.0392 (0.0206)* 0.0824 (0.0293)*** 0.1268 (0.0295)*** 0.0054 (0.0022)** 0.2237 (0.0341)*** 0.0015 (0.0105) 0.0057 (0.0066) 0.0055 (0.0082) 0.2613 (0.0651)*** ‐0.0763 (0.0668) ‐0.1578 (0.1335) ‐0.2053
‐0.0041 (0.0169) ‐0.0361 (0.0123)*** ‐0.0046 (0.0236) 0.0062 (0.0072) ‐0.2629 (0.1119)** 0.1781 (0.0926)* ‐0.0054 (0.0276) 0.0330 (0.0206) 0.0792 (0.0290)*** 0.1375 (0.0330)*** 0.0038 (0.0022)* 0.2238 (0.0366)*** 0.0028 (0.0100) 0.0042 (0.0066) 0.0108 (0.0082) 0.2172 (0.0566)*** ‐0.0887 (0.0678) ‐0.1831 (0.1356) ‐0.2415
‐0.0110 (0.0195) ‐0.0325 (0.0140)** ‐0.0003 (0.0244) 0.0060 (0.0074) ‐0.3643 (0.1438)** 0.1762 (0.0968)* ‐0.0036 (0.0287) 0.0348 (0.0217) 0.0827 (0.0300)*** 0.1333 (0.0337)*** 0.0043 (0.0022)* 0.2196 (0.0375)*** 0.0068 (0.0103) 0.0051 (0.0069) 0.0100 (0.0086) 0.2592 (0.0713)*** ‐0.1032 (0.0711) ‐0.2114 (0.1422) ‐0.2844
‐0.0084 (0.0181) ‐0.0369 (0.0160)** ‐0.9425 (0.0375)*** ‐0.0033 (0.0093) ‐0.3213 (0.2226) 0.1794 (0.1330) ‐0.0069 (0.0365) 0.0326 (0.0280) 0.0829 (0.0379)** 0.1354 (0.0469)*** 0.0071 (0.0025)*** 0.2387 (0.0582)*** 0.0085 (0.0120) 0.0063 (0.0086) 0.0017 (0.0110) 0.2185 (0.1014)** ‐0.0089 (0.0802) ‐0.0239 (0.1594) ‐0.0080
‐0.0152 (0.0193) ‐0.0346 (0.0150)** ‐0.9639 (0.0327)*** 0.0012 (0.0080) ‐0.3571 (0.1644)** 0.1955 (0.1118)* ‐0.0068 (0.0298) 0.0223 (0.0224) 0.0751 (0.0309)** 0.1298 (0.0350)*** 0.0036 (0.0023) 0.2439 (0.0428)*** 0.0054 (0.0102) 0.0005 (0.0074) 0.0121 (0.0090) 0.2536 (0.0831)*** ‐0.0835 (0.0714) ‐0.1738 (0.1428) ‐0.2277
(0.2206) (0.2222) (0.2098) (0.2384) (0.2000) (0.2030) (0.2130) (0.2386) ‐0.1579 ‐0.1660 ‐0.2117 ‐0.0684 ‐0.2770 ‐0.3294 ‐0.3839 ‐0.0066 (0.2943) (0.2966) (0.2800) (0.3181) (0.2667) (0.2707) (0.2841) (0.3182) Year 2006 ‐0.1700 ‐0.1828 ‐0.2401 ‐0.0623 ‐0.3228 ‐0.3853 ‐0.4553 0.0028 (0.3681) (0.3710) (0.3501) (0.3978) (0.3336) (0.3385) (0.3553) (0.3982) Year 2008 ‐0.1939 ‐0.2110 ‐0.2789 ‐0.0646 ‐0.3801 ‐0.4536 ‐0.5395 ‐0.0031 (0.4792) (0.4426) (0.4462) (0.4208) (0.4785) (0.4008) (0.4066) (0.4268) MSA dummy x year ‐0.0063 ‐0.0049 ‐0.0057 ‐0.0051 ‐0.0051 ‐0.0057 ‐0.0058 ‐0.0047 (0.0018)*** (0.0018)*** (0.0017)*** (0.0019)*** (0.0017)*** (0.0017)*** (0.0018)*** (0.0019)** Log population in 2000 ‐0.0006 ‐0.0019 ‐0.0010 ‐0.0016 ‐0.0015 ‐0.0017 ‐0.0018 ‐0.0012 census x year (0.0007) (0.0007)*** (0.0007) (0.0007)** (0.0006)** (0.0006)*** (0.0007)*** (0.0007)* ‐0.0088 ‐0.0092 ‐0.0092 ‐0.0063 ‐0.0065 ‐0.0059 ‐0.0093 % Hispanic in 2000 census ‐0.0098 (0.0054)* (0.0055) (0.0053)* (0.0057) (0.0050) (0.0050) (0.0053) (0.0058) x year % Black in 2000 census ‐0.0150 ‐0.0142 ‐0.0157 ‐0.0150 ‐0.0125 ‐0.0118 ‐0.0098 ‐0.0155 x year (0.0046)*** (0.0045)*** (0.0045)*** (0.0049)*** (0.0043)*** (0.0044)*** (0.0045)** (0.0049)*** ‐0.0408 ‐0.0369 ‐0.0469 % age 65+ in 2000 census ‐0.0473 ‐0.0317 ‐0.0411 ‐0.0398 ‐0.0375 (0.0200)** (0.0206) (0.0194)** (0.0220)* (0.0187)** (0.0191)** (0.0198)* (0.0228)** x year % age 25‐64 in 2000 ‐0.0303 ‐0.0361 ‐0.0281 ‐0.0391 ‐0.0314 ‐0.0145 ‐0.0150 ‐0.0368 census x year (0.0157)* (0.0154)** (0.0154)* (0.0158)** (0.0140)** (0.0145) (0.0150) (0.0161)** ‐0.0067 % university education in ‐0.0089 ‐0.0104 ‐0.0119 ‐0.0066 ‐0.0083 ‐0.0065 ‐0.0047 (0.0150) (0.0154) (0.0139) (0.0166) (0.0133) (0.0137) (0.0144) (0.0168) 2000 census x year Log median home value 0.0097 0.0121 0.0107 0.0103 0.0123 0.0125 0.0131 0.0093 in 2000 census x year (0.0028)*** (0.0027)*** (0.0027)*** (0.0030)*** (0.0025)*** (0.0025)*** (0.0026)*** (0.0030)*** Log median hh income in 0.0015 0.0006 0.0013 0.0015 0.0009 0.0001 0.0002 0.0012 2000 census x year (0.0048) (0.0046) (0.0045) (0.0050) (0.0044) (0.0043) (0.0046) (0.0051) Constant 15.2279 14.7670 15.9504 15.3689 15.9295 16.1173 16.1412 15.4592 (0.4685)*** (0.6778)*** (0.8226)*** (0.5672)*** (0.6478)*** (0.6382)*** (0.6425)*** (0.7763)*** Unit of observation is a hospital‐year. Data biannually from 1996 to 2008. Regressions include hospital‐specific fixed effects, differenced out at means. Year 2004
(0.2139) ‐0.2983 (0.2852) ‐0.3607 (0.3568) ‐0.4429 (0.4286) ‐0.0057 (0.0018)*** ‐0.0014 (0.0007)** ‐0.0067 (0.0053) ‐0.0106 (0.0045)** ‐0.0442 (0.0202)** ‐0.0127 (0.0152) ‐0.0053 (0.0146) 0.0126 (0.0026)*** ‐0.0008 (0.0047) 15.4361 (0.8327)***
Table 5: Main effects by technology, by years since adoption
(1) Log total costs
(2) Log total costs
Technology
Basic EMR adoption 0.0170 (0.0067)** ‐0.0042 (0.0083) 9122 2178 0.80 ‐0.2988 (0.0899)*** ‐0.1385 (0.0626)** 0.0201 (0.0055)*** 0.0105 (0.0040)*** ‐0.0038 (0.0067) 0.0849 (0.0180)*** 0.0008 (0.0078) 0.0185 (0.0080)** ‐0.0048 (0.0184) ‐0.0337 (0.0151)** 0.0132 (0.0264) 0.0032 (0.0084) ‐0.3804 (0.1857)** 0.1205 (0.1158) ‐0.0220 (0.0370) 0.0180 (0.0275) 0.0700 (0.0372)* 0.1371 (0.0446)*** 0.0075
Advanced EMR adoption 0.0310 (0.0075)*** 0.0112 (0.0097)
(3) Log total costs per admit Basic EMR adoption 0.0159 (0.0069)** ‐0.0054 (0.0085)
13193 3175 0.79
7687 1564 0.78
‐0.3961 (0.0793)*** ‐0.1614 (0.0619)*** 0.0193 (0.0041)*** 0.0071 (0.0015)*** 0.0061 (0.0051) 0.0905 (0.0166)*** 0.0045 (0.0071) 0.0091 (0.0076) ‐0.0115 (0.0196) ‐0.0320 (0.0139)** ‐0.0006 (0.0244) 0.0062 (0.0074) ‐0.3646 (0.1436)** 0.1763 (0.0968)* ‐0.0041 (0.0287) 0.0351 (0.0218) 0.0837 (0.0301)*** 0.1332 (0.0339)*** 0.0044
‐0.4804 (0.1345)*** ‐0.0806 (0.0890) 0.0299 (0.0082)*** 0.0100 (0.0048)** ‐0.0082 (0.0090) 0.0974 (0.0192)*** 0.0029 (0.0079) 0.0163 (0.0079)** ‐0.0084 (0.0181) ‐0.0371 (0.0160)** ‐0.9422 (0.0375)*** ‐0.0033 (0.0093) ‐0.3211 (0.2225) 0.1817 (0.1331) ‐0.0073 (0.0365) 0.0323 (0.0279) 0.0828 (0.0379)** 0.1353 (0.0470)*** 0.0070
Adopt in previous 2 year period Adopt at least 2 years earlier Observations # of hospitals R‐squared CONTROLS Log inpatient days Log outpatient visits Log inpatient days x Log inpatient days Log outpatient visits x Log outpatient visits Log inpatient days x Log outpatient visits Log total hospital beds Independent practice association hospital Management service organization hospital Equity model hospital Foundation hospital Log admissions Births (000s) Full time physicians and dentists (000s) Percent births For‐profit ownership Non‐secular nonprofit ownership Non‐profit church ownership Log number of discharges Medicare Log number of discharges
(4) Log total costs per admit Advanced EMR adoption 0.0336 (0.0078)*** 0.0151 (0.0099) 11082 2268 0.77 ‐0.4497 (0.0920)*** ‐0.0837 (0.0897) 0.0238 (0.0047)*** 0.0065 (0.0015)*** 0.0001 (0.0072) 0.1046 (0.0171)*** 0.0091 (0.0072) 0.0082 (0.0076) ‐0.0156 (0.0193) ‐0.0340 (0.0149)** ‐0.9642 (0.0327)*** 0.0013 (0.0080) ‐0.3574 (0.1643)** 0.1958 (0.1118)* ‐0.0074 (0.0298) 0.0224 (0.0224) 0.0758 (0.0310)** 0.1298 (0.0351)*** 0.0037
Medicaid Log number of discharges total
(0.0025)*** (0.0022)** (0.0025)*** (0.0023) 0.2337 0.2198 0.2390 0.2440 (0.0519)*** (0.0376)*** (0.0582)*** (0.0429)*** Residency or Member of Council 0.0081 0.0073 0.0085 0.0057 Teaching Hospitals (0.0119) (0.0104) (0.0120) (0.0102) Vertically integrated with doctors 0.0110 0.0052 0.0064 0.0007 (0.0082) (0.0069) (0.0086) (0.0074) Vert. integ. w drs (excl. integrated ‐0.0019 0.0101 0.0016 0.0122 salary model) (0.0105) (0.0086) (0.0109) (0.0090) 0.2539 FT physicians / total hospital beds 0.2485 0.2597 0.2184 (0.0817)*** (0.0714)*** (0.1013)** (0.0832)*** Year 1998 ‐0.0219 ‐0.1021 ‐0.0075 ‐0.0824 (0.0801) (0.0711) (0.0802) (0.0714) Year 2000 ‐0.0496 ‐0.2094 ‐0.0221 ‐0.1717 (0.1595) (0.1422) (0.1596) (0.1428) Year 2002 ‐0.0465 ‐0.2823 ‐0.0053 ‐0.2255 (0.2388) (0.2129) (0.2389) (0.2137) Year 2004 ‐0.0675 ‐0.3805 ‐0.0030 ‐0.2951 (0.3187) (0.2840) (0.3187) (0.2851) Year 2006 ‐0.0615 ‐0.4510 0.0072 ‐0.3567 (0.3986) (0.3551) (0.3988) (0.3566) Year 2008 ‐0.0636 ‐0.5346 0.0021 ‐0.4387 (0.4794) (0.4266) (0.4800) (0.4283) MSA dummy x year ‐0.0051 ‐0.0059 ‐0.0047 ‐0.0057 (0.0019)*** (0.0018)*** (0.0019)** (0.0018)*** Log population in 2000 census x ‐0.0016 ‐0.0018 ‐0.0012 ‐0.0014 year (0.0007)** (0.0007)*** (0.0007) (0.0007)** % Hispanic in 2000 census ‐0.0093 ‐0.0059 ‐0.0094 ‐0.0067 x year (0.0057) (0.0053) (0.0058) (0.0053) % Black in 2000 census ‐0.0154 ‐0.0102 ‐0.0158 ‐0.0109 x year (0.0049)*** (0.0045)** (0.0049)*** (0.0045)** ‐0.0402 ‐0.0370 ‐0.0473 ‐0.0444 % age 65+ in 2000 census x year (0.0221)* (0.0199)* (0.0228)** (0.0203)** % age 25‐64 in 2000 census x year ‐0.0388 ‐0.0151 ‐0.0366 ‐0.0129 (0.0158)** (0.0150) (0.0160)** (0.0151) % university education in 2000 ‐0.0065 ‐0.0043 ‐0.0066 ‐0.0050 census x year (0.0166) (0.0143) (0.0168) (0.0145) 0.0102 0.0130 0.0093 0.0125 Log median home value in 2000 (0.0030)*** (0.0026)*** (0.0030)*** (0.0026)*** census x year Log median hh income in 2000 0.0015 0.0002 0.0012 ‐0.0007 census x year (0.0050) (0.0046) (0.0051) (0.0047) Constant 15.3830 16.1381 15.4783 15.4264 (0.5693)*** (0.6436)*** (0.7786)*** (0.8342)*** Unit of observation is a hospital‐year. Data biannually from 1996 to 2008. Regressions include hospital‐specific fixed effects, differenced out at means.
Table 6: Interactions with IT‐intensive location (1) (2) (3) (4) Log total costs Definition of IT‐intensive Top quartile county IT MSA location intensive industries Technology
Basic EMR adoption
Adopt in previous 2 year 0.0241 (0.0088)*** period Adopt at least 2 years 0.0188 earlier (0.0110)* Adopt in previous 2 yr pd ‐0.0212 x IT‐intensive location (0.0134) Adopt at least 2 yrs earlier ‐0.0574 x IT‐intensive location (0.0166)*** Observations 8778 # of hospitals 2062 R‐squared 0.81 CONTROLS IT‐intensive location x 0.0035 year (0.0021)* Log inpatient days ‐0.4044 (0.0954)*** Log outpatient visits ‐0.1085 (0.0709) Log inpatient days x Log 0.0270 inpatient days (0.0056)*** Log outpatient visits x Log 0.0103 outpatient visits (0.0043)** Log inpatient days x Log ‐0.0064 (0.0073) outpatient visits Log total hospital beds 0.0849 (0.0177)*** Independent practice 0.0007
Advanced EMR adoption 0.0336 (0.0114)*** 0.0331 (0.0145)** ‐0.0068 (0.0152) ‐0.0388 (0.0193)** 12778 3033 0.79 ‐0.0003 (0.0015) ‐0.4792 (0.0847)*** ‐0.1593 (0.0702)** 0.0235 (0.0037)*** 0.0071 (0.0015)*** 0.0060 (0.0058) 0.0921 (0.0170)*** 0.0052
0.0201 (0.0101)** 0.0153 (0.0129) ‐0.0054 (0.0132) ‐0.0312 (0.0166)*
(9) (10) Log total costs per admit % of all local firms that % of all local firms that Top quartile county IT had adopted basic IT by had adopted advanced IT using industries 2000 by 2000 Advanced Basic EMR Advanced Basic EMR Advanced Basic EMR Advanced EMR adoption EMR adoption EMR adoption EMR adoption adoption adoption adoption 0.0304 0.0991 0.1381 0.0234 0.0500 0.0245 0.0375 (0.0124)** (0.0380)*** (0.0525)*** (0.0113)** (0.0151)*** (0.0091)*** (0.0119)*** 0.0551 0.0732 0.1244 0.0107 0.0426 0.0221 0.0383 (0.0181)*** (0.0464) (0.0590)** (0.0131) (0.0199)** (0.0113)* (0.0149)** 0.0009 ‐0.1085 ‐0.1353 ‐0.0891 ‐0.1802 ‐0.0217 ‐0.0091 (0.0154) (0.0467)** (0.0622)** (0.0857) (0.1039)* (0.0137) (0.0156) ‐0.0618 ‐0.1041 ‐0.1390 ‐0.1856 ‐0.2744 ‐0.0626 ‐0.0391 (0.0214)*** (0.0577)* (0.0705)** (0.0901)** (0.1418)* (0.0169)*** (0.0195)**
9122 2178 0.80
13193 3175 0.79
8778 2062 0.81
12778 3033 0.79
8778 2062 0.81
12778 3033 0.79
7577 1539 0.78
10958 2240 0.77
‐0.0032 (0.0022) ‐0.3020 (0.0899)*** ‐0.1368 (0.0630)** 0.0203 (0.0055)*** 0.0104 (0.0040)*** ‐0.0038 (0.0067) 0.0849 (0.0180)*** 0.0004
‐0.0049 (0.0018)*** ‐0.3956 (0.0791)*** ‐0.1621 (0.0618)*** 0.0192 (0.0041)*** 0.0072 (0.0015)*** 0.0061 (0.0051) 0.0915 (0.0166)*** 0.0040
0.0046 (0.0042) ‐0.4092 (0.0949)*** ‐0.1062 (0.0707) 0.0272 (0.0056)*** 0.0102 (0.0042)** ‐0.0064 (0.0072) 0.0838 (0.0178)*** 0.0007
0.0011 (0.0028) ‐0.4781 (0.0844)*** ‐0.1583 (0.0701)** 0.0235 (0.0037)*** 0.0071 (0.0015)*** 0.0059 (0.0058) 0.0918 (0.0169)*** 0.0053
‐0.0030 (0.0089) ‐0.4008 (0.0948)*** ‐0.1114 (0.0705) 0.0268 (0.0056)*** 0.0104 (0.0042)** ‐0.0064 (0.0072) 0.0842 (0.0177)*** 0.0008
‐0.0038 (0.0067) ‐0.4765 (0.0846)*** ‐0.1594 (0.0703)** 0.0234 (0.0037)*** 0.0071 (0.0015)*** 0.0059 (0.0058) 0.0922 (0.0170)*** 0.0053
0.0040 (0.0021)* ‐0.5269 (0.1317)*** ‐0.0715 (0.0888) 0.0339 (0.0082)*** 0.0104 (0.0049)** ‐0.0102 (0.0090) 0.0976 (0.0184)*** 0.0010
‐0.0002 (0.0015) ‐0.4787 (0.0906)*** ‐0.0895 (0.0908) 0.0253 (0.0045)*** 0.0066 (0.0015)*** 0.0004 (0.0073) 0.1044 (0.0168)*** 0.0083
Basic EMR adoption
(5)
(6)
(7)
(8)
association hospital Management service organization hospital Equity model hospital Foundation hospital Log admissions Births (000s) Full time physicians and dentists (000s) Percent births For‐profit ownership Non‐secular nonprofit ownership Non‐profit church ownership Log number of discharges Medicare Log number of discharges Medicaid Log number of discharges total Residency or Member of Council Teaching Hosps Vertically integrated with doctors Vert. integ. w drs (excl. integrated salary mdl) FT physicians / total hospital beds Year 1998
(0.0079) 0.0197 (0.0079)** ‐0.0038 (0.0179) ‐0.0333 (0.0150)** 0.0131 (0.0269) ‐0.0015 (0.0083) ‐0.3418 (0.1847)* 0.1259 (0.1165) 0.0003 (0.0355) 0.0214 (0.0276) 0.0745 (0.0372)** 0.1306 (0.0443)*** 0.0069 (0.0026)*** 0.2620 (0.0555)*** 0.0110 (0.0115) 0.0107 (0.0082) ‐0.0013 (0.0103) 0.2274 (0.0813)*** ‐0.0053 (0.0794)
(0.0072) 0.0092 (0.0077) ‐0.0102 (0.0195) ‐0.0320 (0.0140)** 0.0006 (0.0249) 0.0034 (0.0073) ‐0.3431 (0.1430)** 0.1684 (0.0978)* 0.0070 (0.0286) 0.0366 (0.0220)* 0.0858 (0.0304)*** 0.1274 (0.0335)*** 0.0037 (0.0022)* 0.2374 (0.0387)*** 0.0088 (0.0101) 0.0043 (0.0070) 0.0113 (0.0086) 0.2471 (0.0713)*** ‐0.0914 (0.0719)
(0.0078) 0.0185 (0.0080)** ‐0.0055 (0.0182) ‐0.0336 (0.0150)** 0.0123 (0.0264) 0.0032 (0.0083) ‐0.3767 (0.1850)** 0.1224 (0.1156) ‐0.0219 (0.0371) 0.0185 (0.0276) 0.0704 (0.0373)* 0.1373 (0.0443)*** 0.0074 (0.0025)*** 0.2341 (0.0518)*** 0.0082 (0.0119) 0.0113 (0.0082) ‐0.0018 (0.0105) 0.2467 (0.0814)*** ‐0.0232 (0.0802)
(0.0071) 0.0088 (0.0076) ‐0.0117 (0.0194) ‐0.0316 (0.0139)** ‐0.0007 (0.0244) 0.0065 (0.0074) ‐0.3651 (0.1432)** 0.1721 (0.0968)* ‐0.0041 (0.0287) 0.0347 (0.0217) 0.0838 (0.0301)*** 0.1341 (0.0337)*** 0.0044 (0.0022)* 0.2196 (0.0375)*** 0.0080 (0.0103) 0.0053 (0.0069) 0.0104 (0.0086) 0.2599 (0.0714)*** ‐0.1003 (0.0710)
(0.0079) 0.0196 (0.0079)** ‐0.0048 (0.0181) ‐0.0342 (0.0151)** 0.0121 (0.0267) ‐0.0011 (0.0083) ‐0.3456 (0.1834)* 0.1257 (0.1165) 0.0004 (0.0355) 0.0209 (0.0273) 0.0734 (0.0370)** 0.1309 (0.0439)*** 0.0069 (0.0026)*** 0.2618 (0.0554)*** 0.0101 (0.0114) 0.0103 (0.0082) ‐0.0010 (0.0104) 0.2306 (0.0802)*** ‐0.0101 (0.0789)
(0.0072) 0.0090 (0.0077) ‐0.0104 (0.0193) ‐0.0322 (0.0140)** ‐0.0002 (0.0248) 0.0031 (0.0073) ‐0.3440 (0.1426)** 0.1703 (0.0976)* 0.0074 (0.0286) 0.0371 (0.0219)* 0.0858 (0.0304)*** 0.1274 (0.0333)*** 0.0038 (0.0022)* 0.2374 (0.0386)*** 0.0086 (0.0101) 0.0044 (0.0070) 0.0111 (0.0086) 0.2492 (0.0703)*** ‐0.0899 (0.0706)
(0.0079) 0.0196 (0.0080)** ‐0.0021 (0.0180) ‐0.0347 (0.0150)** 0.0114 (0.0268) ‐0.0009 (0.0083) ‐0.3400 (0.1837)* 0.1261 (0.1162) ‐0.0032 (0.0358) 0.0191 (0.0276) 0.0734 (0.0373)** 0.1314 (0.0437)*** 0.0068 (0.0026)*** 0.2616 (0.0552)*** 0.0104 (0.0114) 0.0105 (0.0082) ‐0.0007 (0.0104) 0.2277 (0.0807)*** ‐0.0025 (0.0790)
(0.0072) 0.0097 (0.0077) ‐0.0086 (0.0193) ‐0.0331 (0.0139)** ‐0.0006 (0.0249) 0.0035 (0.0073) ‐0.3413 (0.1429)** 0.1680 (0.0978)* 0.0070 (0.0287) 0.0370 (0.0220)* 0.0864 (0.0304)*** 0.1281 (0.0332)*** 0.0038 (0.0022)* 0.2369 (0.0385)*** 0.0087 (0.0101) 0.0045 (0.0070) 0.0112 (0.0086) 0.2475 (0.0708)*** ‐0.0891 (0.0709)
(0.0079) 0.0179 (0.0078)** ‐0.0077 (0.0175) ‐0.0338 (0.0158)** ‐0.9415 (0.0373)*** ‐0.0074 (0.0092) ‐0.3087 (0.2229) 0.1878 (0.1300) 0.0036 (0.0348) 0.0305 (0.0277) 0.0820 (0.0375)** 0.1259 (0.0453)*** 0.0065 (0.0025)** 0.2578 (0.0599)*** 0.0114 (0.0117) 0.0066 (0.0084) 0.0019 (0.0107) 0.2098 (0.1015)** 0.0060 (0.0795)
(0.0073) 0.0087 (0.0076) ‐0.0144 (0.0193) ‐0.0321 (0.0148)** ‐0.9626 (0.0329)*** ‐0.0008 (0.0080) ‐0.3441 (0.1638)** 0.1905 (0.1113)* ‐0.0030 (0.0295) 0.0202 (0.0224) 0.0748 (0.0309)** 0.1220 (0.0338)*** 0.0030 (0.0023) 0.2559 (0.0430)*** 0.0074 (0.0100) 0.0001 (0.0073) 0.0133 (0.0089) 0.2464 (0.0830)*** ‐0.0742 (0.0723)
Year 2000
‐0.0183 ‐0.1884 ‐0.0518 ‐0.2060 ‐0.0288 ‐0.1853 ‐0.0133 ‐0.1840 0.0041 ‐0.1553 (0.1582) (0.1438) (0.1596) (0.1419) (0.1572) (0.1411) (0.1573) (0.1418) (0.1583) (0.1446) Year 2002 0.0012 ‐0.2504 ‐0.0500 ‐0.2772 ‐0.0144 ‐0.2459 0.0086 ‐0.2441 0.0347 ‐0.2005 (0.2369) (0.2154) (0.2389) (0.2126) (0.2353) (0.2114) (0.2356) (0.2124) (0.2370) (0.2165) Year 2004 ‐0.0026 ‐0.3371 ‐0.0722 ‐0.3735 ‐0.0231 ‐0.3313 0.0074 ‐0.3288 0.0514 ‐0.2611 (0.3161) (0.2873) (0.3188) (0.2835) (0.3140) (0.2819) (0.3142) (0.2832) (0.3161) (0.2888) ‐0.3149 Year 2006 0.0200 ‐0.3966 ‐0.0667 ‐0.4417 ‐0.0057 ‐0.3895 0.0325 ‐0.3862 0.0747 (0.3954) (0.3593) (0.3987) (0.3546) (0.3927) (0.3526) (0.3931) (0.3542) (0.3955) (0.3612) Year 2008 0.0295 ‐0.4740 ‐0.0696 ‐0.5233 ‐0.0022 ‐0.4658 0.0440 ‐0.4616 0.0799 ‐0.3907 (0.4757) (0.4316) (0.4796) (0.4259) (0.4725) (0.4236) (0.4729) (0.4256) (0.4762) (0.4340) MSA dummy x year ‐0.0051 ‐0.0059 Above Above ‐0.0053 ‐0.0062 ‐0.0052 ‐0.0060 ‐0.0046 ‐0.0058 (0.0019)*** (0.0018)*** (0.0020)*** (0.0018)*** (0.0019)*** (0.0018)*** (0.0019)** (0.0018)*** ‐0.0016 ‐0.0018 ‐0.0018 ‐0.0019 ‐0.0018 ‐0.0019 ‐0.0017 ‐0.0015 Log population in 2000 ‐0.0020 ‐0.0019 (0.0008)** (0.0007)*** (0.0007)** (0.0007)*** (0.0008)** (0.0007)*** (0.0007)** (0.0007)*** (0.0008)** (0.0007)** census x year % Hispanic in 2000 census ‐0.0082 ‐0.0061 ‐0.0088 ‐0.0057 ‐0.0085 ‐0.0063 ‐0.0084 ‐0.0059 ‐0.0083 ‐0.0065 x year (0.0057) (0.0053) (0.0057) (0.0053) (0.0058) (0.0053) (0.0056) (0.0053) (0.0058) (0.0053) ‐0.0102 ‐0.0163 ‐0.0108 % Black in 2000 census ‐0.0160 ‐0.0101 ‐0.0153 ‐0.0099 ‐0.0160 ‐0.0103 ‐0.0156 (0.0049)*** (0.0045)** (0.0049)*** (0.0045)** (0.0049)*** (0.0045)** (0.0049)*** (0.0045)** (0.0048)*** (0.0045)** x year % age 65+ in 2000 census ‐0.0405 ‐0.0397 ‐0.0387 ‐0.0369 ‐0.0431 ‐0.0425 ‐0.0425 ‐0.0416 ‐0.0460 ‐0.0444 x year (0.0224)* (0.0203)* (0.0220)* (0.0199)* (0.0221)* (0.0199)** (0.0222)* (0.0200)** (0.0229)** (0.0205)** % age 25‐64 in 2000 ‐0.0399 ‐0.0167 ‐0.0388 ‐0.0153 ‐0.0393 ‐0.0161 ‐0.0405 ‐0.0162 ‐0.0369 ‐0.0141 (0.0158)** (0.0151) (0.0158)** (0.0148) (0.0158)** (0.0150) (0.0159)** (0.0151) (0.0159)** (0.0152) census x year % university education in ‐0.0001 0.0005 ‐0.0053 ‐0.0031 ‐0.0005 ‐0.0017 0.0018 ‐0.0009 ‐0.0002 0.0015 2000 census x year (0.0170) (0.0145) (0.0165) (0.0143) (0.0168) (0.0145) (0.0168) (0.0146) (0.0172) (0.0147) Log median home value 0.0094 0.0125 0.0100 0.0129 0.0094 0.0127 0.0093 0.0125 0.0086 0.0119 in 2000 census x year (0.0030)*** (0.0026)*** (0.0030)*** (0.0026)*** (0.0030)*** (0.0026)*** (0.0030)*** (0.0026)*** (0.0031)*** (0.0026)*** Log median hh income in 0.0019 0.0005 0.0016 0.0002 0.0018 0.0001 0.0020 0.0004 0.0016 ‐0.0003 2000 census x year (0.0050) (0.0046) (0.0050) (0.0045) (0.0050) (0.0046) (0.0050) (0.0046) (0.0051) (0.0047) Constant 15.5781 16.4562 15.3866 16.1276 15.6001 16.4525 15.6004 16.4546 15.5312 15.5346 (0.6426)*** (0.7445)*** (0.5705)*** (0.6430)*** (0.6374)*** (0.7410)*** (0.6360)*** (0.7434)*** (0.7725)*** (0.8451)*** Unit of observation is a hospital‐year. Data biannually from 1996 to 2008. Regressions include hospital‐specific fixed effects, differenced out at means.
Table 7: Interactions with internal HIT experience Definition of HIT experience
(1)
(2)
(3) (4) Log total costs Employ computer Have at least 1 clinical programmers in 1996 application in 1996 Technology Basic EMR Advanced Basic EMR Advanced adoption EMR adoption EMR adoption adoption Adopt in previous 2 year 0.0042 0.0400 0.0020 0.0599 (0.0094) (0.0115)*** (0.0168) (0.0253)** period Adopt at least 2 years ‐0.0169 0.0180 ‐0.0125 0.0289 earlier (0.0108) (0.0160) (0.0194) (0.0335) Adopt in previous 2 year 0.0053 ‐0.0425 0.0007 ‐0.0443 period x HIT experience (0.0224) (0.0211)** (0.0191) (0.0271) Adopt at least 2 years earlier 0.0007 ‐0.0176 ‐0.0069 ‐0.0238 (0.0328) (0.0298) (0.0231) (0.0366) x HIT experience Observations 3976 5581 5550 3966 # of hospitals 816 1162 1156 814 R‐squared 0.84 0.82 0.82 0.84 CONTROLS HIT experience x year ‐0.0034 ‐0.0012 0.0023 0.0012 (0.0034) (0.0024) (0.0018) (0.0024) Log inpatient days ‐0.9459 ‐0.9678 ‐0.9233 ‐0.9722 (0.2977)*** (0.2446)*** (0.3038)*** (0.2447)*** Log outpatient visits ‐0.3104 ‐0.3453 ‐0.3227 ‐0.3419 (0.2060) (0.1563)** (0.2064) (0.1552)** Log inpatient days x Log 0.0560 0.0349 0.0553 0.0351 inpatient days (0.0154)*** (0.0117)*** (0.0156)*** (0.0118)*** Log outpatient visits x Log 0.0250 0.0073 0.0258 0.0071 outpatient visits (0.0105)** (0.0014)*** (0.0105)** (0.0013)*** Log inpatient days x Log ‐0.0198 0.0228 ‐0.0206 0.0227 outpatient visits (0.0165) (0.0134)* (0.0165) (0.0133)* Log total hospital beds 0.1213 0.1005 0.1252 0.1030 (0.0279)*** (0.0263)*** (0.0278)*** (0.0264)*** 0.0087 0.0022 0.0065 Independent practice 0.0044 (0.0102) (0.0097) (0.0101) (0.0096) association hospital Management service 0.0203 0.0082 0.0206 0.0117 organization hospital (0.0098)** (0.0102) (0.0097)** (0.0101) Equity model hospital 0.0211 ‐0.0032 0.0217 ‐0.0058 (0.0327) (0.0321) (0.0323) (0.0319) Foundation hospital ‐0.0441 ‐0.0647 ‐0.0442 ‐0.0629 (0.0151)*** (0.0206)*** (0.0154)*** (0.0205)*** Log admissions 0.0198 ‐0.0551 0.0162 ‐0.0582 (0.0473) (0.0615) (0.0475) (0.0615) Births (000s) ‐0.0141 ‐0.0026 ‐0.0125 ‐0.0014 (0.0181) (0.0120) (0.0180) (0.0120) Full time physicians and ‐0.2257 ‐0.3507 ‐0.2283 ‐0.3478 dentists (000s) (0.2765) (0.2731) (0.2809) (0.2787) Percent births 0.3931 0.2002 0.3573 0.1843 (0.2285)* (0.1720) (0.2274) (0.1730) For‐profit ownership ‐0.0724 ‐0.0258 ‐0.0743 ‐0.0266 (0.0527) (0.0774) (0.0530) (0.0773)
(5) (6) Log total costs per admit Employ computer programmers in 1996 Basic EMR Advanced adoption EMR adoption 0.0030 0.0399 (0.0094) (0.0116)*** ‐0.0180 0.0174 (0.0108)* (0.0162) 0.0060 ‐0.0427 (0.0225) (0.0212)** 0.0013 ‐0.0168 (0.0330) (0.0299) 3899 5451 784 1108 0.80 0.78 ‐0.0035 ‐0.0012 (0.0034) (0.0024) ‐0.9725 ‐0.9625 (0.3160)*** (0.2676)*** ‐0.2913 ‐0.3507 (0.2150) (0.1648)** 0.0581 0.0343 (0.0176)*** (0.0133)** 0.0250 0.0072 (0.0106)** (0.0013)*** ‐0.0216 0.0234 (0.0177) (0.0142)* 0.1179 0.0965 (0.0283)*** (0.0268)*** 0.0054 0.0096 (0.0102) (0.0096) 0.0212 0.0083 (0.0098)** (0.0103) 0.0211 ‐0.0035 (0.0324) (0.0320) ‐0.0439 ‐0.0647 (0.0152)*** (0.0209)*** ‐0.9583 ‐1.0434 (0.0619)*** (0.0479)*** ‐0.0187 ‐0.0049 (0.0183) (0.0119) ‐0.2067 ‐0.3214 (0.2895) (0.2858) 0.4594 0.2445 (0.2341)* (0.1731) ‐0.0715 ‐0.0260 (0.0766) (0.0524)
Non‐secular nonprofit ownership Non‐profit church ownership Log number of discharges Medicare Log number of discharges Medicaid Log number of discharges total Residency or Member of Council Teaching Hospitals Vertically integrated with doctors Vert. integ. w drs (excl. integrated salary model) FT physicians / total hospital beds Year 1998
0.0180 0.0172 0.0166 0.0178 0.0188 0.0176 (0.0669) (0.0451) (0.0673) (0.0454) (0.0660) (0.0448) 0.0049 0.0443 0.0063 0.0464 0.0021 0.0425 (0.0721) (0.0517) (0.0722) (0.0520) (0.0712) (0.0516) 0.3078 0.3105 0.3073 0.3108 0.3000 0.3074 (0.0415)*** (0.0360)*** (0.0416)*** (0.0360)*** (0.0418)*** (0.0370)*** 0.0080 0.0040 0.0093 0.0054 0.0073 0.0037 (0.0033)** (0.0030) (0.0031)*** (0.0030)* (0.0032)** (0.0030) 0.1857 0.2091 0.1840 0.2078 0.1820 0.2062 (0.0439)*** (0.0384)*** (0.0439)*** (0.0383)*** (0.0439)*** (0.0388)*** 0.0009 ‐0.0058 0.0012 ‐0.0052 0.0014 ‐0.0055 (0.0125) (0.0115) (0.0125) (0.0115) (0.0126) (0.0115) 0.0182 0.0124 0.0153 0.0095 0.0180 0.0128 (0.0105)* (0.0095) (0.0104) (0.0093) (0.0105)* (0.0095) ‐0.0123 0.0036 ‐0.0103 0.0054 ‐0.0128 0.0025 (0.0123) (0.0111) (0.0121) (0.0111) (0.0124) (0.0113) 0.2222 0.2690 0.2245 0.2703 0.2143 0.2551 (0.1258)* (0.1262)** (0.1266)* (0.1275)** (0.1327) (0.1333)* 0.1132 ‐0.0018 0.0843 ‐0.0235 0.1143 ‐0.0046 (0.1076) (0.1057) (0.1067) (0.1056) (0.1080) (0.1059) Year 2000 0.2097 ‐0.0127 0.1572 ‐0.0515 0.2109 ‐0.0184 (0.2141) (0.2113) (0.2123) (0.2112) (0.2148) (0.2118) Year 2002 0.3481 0.0066 0.2699 ‐0.0509 0.3499 ‐0.0021 (0.3206) (0.3165) (0.3177) (0.3162) (0.3216) (0.3172) Year 2004 0.4724 0.0176 0.3677 ‐0.0602 0.4742 0.0054 (0.4224) (0.4239) (0.4221) (0.4290) (0.4233) (0.4277) Year 2006 0.6014 0.0355 0.4706 ‐0.0615 0.6033 0.0203 (0.5345) (0.5279) (0.5297) (0.5276) (0.5361) (0.5292) Year 2008 0.7259 0.0386 0.5689 ‐0.0767 0.7273 0.0186 (0.6427) (0.6340) (0.6370) (0.6336) (0.6447) (0.6355) MSA dummy x year ‐0.0056 ‐0.0066 ‐0.0060 ‐0.0066 ‐0.0053 ‐0.0064 (0.0023)** (0.0023)*** (0.0024)** (0.0023)*** (0.0024)** (0.0023)*** Log population in 2000 ‐0.0015 ‐0.0011 ‐0.0017 ‐0.0013 ‐0.0014 ‐0.0011 (0.0010) (0.0010) (0.0010)* (0.0009) (0.0010) (0.0010) census x year % Hispanic in 2000 census ‐0.0112 ‐0.0045 ‐0.0105 ‐0.0048 ‐0.0108 ‐0.0041 x year (0.0078) (0.0079) (0.0078) (0.0080) (0.0078) (0.0079) % Black in 2000 census ‐0.0160 ‐0.0116 ‐0.0163 ‐0.0118 ‐0.0163 ‐0.0120 x year (0.0070)** (0.0067)* (0.0070)** (0.0067)* (0.0069)** (0.0067)* 0.0039 % age 65+ in 2000 census x ‐0.0424 0.0051 ‐0.0455 0.0007 ‐0.0444 (0.0287) (0.0301) (0.0285) (0.0299) (0.0286) (0.0301) year % age 25‐64 in 2000 census ‐0.0043 ‐0.0070 ‐0.0064 ‐0.0092 ‐0.0057 ‐0.0078 x year (0.0204) (0.0232) (0.0200) (0.0229) (0.0204) (0.0233) % university education in 0.0091 ‐0.0057 0.0041 ‐0.0108 0.0078 ‐0.0071 2000 census x year (0.0218) (0.0198) (0.0214) (0.0197) (0.0219) (0.0198) 0.0108 0.0077 0.0113 0.0071 0.0108 Log median home value in 0.0073 (0.0039)* (0.0037)*** (0.0039)** (0.0038)*** (0.0040)* (0.0038)*** 2000 census x year Log median hh income in ‐0.0048 ‐0.0040 ‐0.0036 ‐0.0033 ‐0.0046 ‐0.0038 2000 census x year (0.0068) (0.0074) (0.0067) (0.0074) (0.0068) (0.0074) Constant 19.3883 20.0020 20.0309 19.3676 19.0042 19.6471 (1.6507)*** (1.7325)*** (1.7203)*** (1.6923)*** (1.6232)*** (1.8005)*** Unit of observation is a hospital‐year. Data biannually from 1996 to 2008. Regressions include hospital‐specific fixed effects, differenced out at means.
Table 8: Interactions of adoption with top quartile IT‐using county and with programmers in 1996 Technology
(1) Basic EMR adoption
Adopt in previous 2 year period
0.0140 (0.0113) 0.0054 (0.0149) ‐0.0155 (0.0167) ‐0.0407 (0.0208)* 0.0040 (0.0224) 0.0040 (0.0325)
(2) Advanced EMR adoption 0.0376 (0.0158)** 0.0379 (0.0214)* 0.0087 (0.0206) ‐0.0270 (0.0272) ‐0.0475 (0.0217)** ‐0.0167 (0.0294)
3936 805 0.85
5539 1150 0.82
3936 805 0.85
‐0.0036 (0.0034) 0.0015 (0.0025)
‐0.0010 (0.0024) ‐0.0017 (0.0018)
‐1.0341 (0.2548)*** ‐0.2336 (0.1892) 0.0610 (0.0141)*** 0.0222 (0.0098)** ‐0.0207 (0.0165) 0.1140 (0.0262)*** 0.0054 (0.0103) 0.0193 (0.0098)** 0.0218 (0.0324) ‐0.0418 (0.0150)***
‐1.0137 (0.2285)*** ‐0.3332 (0.1612)** 0.0379 (0.0108)*** 0.0073 (0.0013)*** 0.0218 (0.0138) 0.0980 (0.0260)*** 0.0095 (0.0097) 0.0075 (0.0102) ‐0.0018 (0.0325) ‐0.0636 (0.0204)***
‐0.0072 (0.0072) 0.0006 (0.0025) 0.0058 (0.0077) ‐1.0317 (0.2532)*** ‐0.2313 (0.1891) 0.0614 (0.0140)*** 0.0224 (0.0097)** ‐0.0215 (0.0164) 0.1135 (0.0259)*** 0.0055 (0.0103) 0.0189 (0.0099)* 0.0239 (0.0325) ‐0.0415 (0.0149)***
Adopt at least 2 years earlier Adopt in previous 2 year period x IT intensive county Adopt at least 2 years earlier x IT intensive county Adopt in previous 2 year period x Programmer 1996 Adopt at least 2 years earlier x Programmer 1996 Adopt in previous 2 year period x IT intensive county x Programmer 1996 Adopt at least 2 years earlier x IT intensive county x Programmer 1996 Observations # of hospitals R‐squared CONTROLS Top quartile IT using county x year Programmer in 1996 x year Top quartile IT using county x Programmer x year Log inpatient days Log outpatient visits Log inpatient days x Log inpatient days Log outpatient visits x Log outpatient visits Log inpatient days x Log outpatient visits Log total hospital beds Independent practice association hospital Management service organization hospital Equity model hospital Foundation hospital
(3) Basic EMR adoption 0.0052 (0.0117) ‐0.0061 (0.0150) 0.0016 (0.0186) ‐0.0181 (0.0211) 0.0573 (0.0332)* 0.0765 (0.0573) ‐0.0845 (0.0436)* ‐0.1137 (0.0681)*
(4) Advanced EMR adoption 0.0301 (0.0171)* 0.0336 (0.0233) 0.0234 (0.0238) ‐0.0195 (0.0314) 0.0012 (0.0412) 0.0076 (0.0499) ‐0.0700 (0.0487) ‐0.0336 (0.0610) 5539 1150 0.82 ‐0.0012 (0.0042) ‐0.0017 (0.0018) 0.0003 (0.0050) ‐1.0096 (0.2281)*** ‐0.3354 (0.1615)** 0.0376 (0.0108)*** 0.0073 (0.0013)*** 0.0219 (0.0138) 0.0983 (0.0258)*** 0.0091 (0.0097) 0.0077 (0.0102) ‐0.0013 (0.0327) ‐0.0639 (0.0204)***
Log admissions
0.0340 (0.0590) Births (000s) ‐0.0212 (0.0165) Full time physicians and dentists (000s) ‐0.1994 (0.2724) Percent births 0.4145 (0.2156)* For‐profit ownership ‐0.0463 (0.0747) Non‐secular nonprofit ownership 0.0156 (0.0672) Non‐profit church ownership 0.0064 (0.0721) Log number of discharges Medicare 0.2833 (0.0362)*** Log number of discharges Medicaid 0.0072 (0.0033)** Log number of discharges total 0.2176 (0.0372)*** Residency or Member of Council Teaching 0.0020 Hospitals (0.0125) Vertically integrated with doctors 0.0168 (0.0106) Vert. integ. w drs (excl. integrated salary ‐0.0127 model) (0.0124) FT physicians / total hospital beds 0.2106 (0.1245)* Year 1998 0.1133 (0.1073) Year 2000 0.2102 (0.2136) Year 2002 0.3503 (0.3199) Year 2004 0.4759 (0.4268) Year 2006 0.6061 (0.5334) Year 2008 0.7283 (0.6413) MSA dummy x year ‐0.0056 (0.0023)** ‐0.0014 Log population in 2000 census x year (0.0010) % Hispanic in 2000 census ‐0.0122 x year (0.0078) % Black in 2000 census ‐0.0165 x year (0.0069)** % age 65+ in 2000 census x year ‐0.0447 (0.0281) ‐0.0078 % age 25‐64 in 2000 census x year (0.0212)
‐0.0506 (0.0462) ‐0.0054 (0.0111) ‐0.3355 (0.2692) 0.2061 (0.1661) ‐0.0111 (0.0511) 0.0151 (0.0449) 0.0458 (0.0516) 0.2911 (0.0322)*** 0.0036 (0.0030) 0.2323 (0.0343)*** ‐0.0055 (0.0114) 0.0117 (0.0095) 0.0037 (0.0112) 0.2617 (0.1251)** ‐0.0110 (0.1060) ‐0.0310 (0.2121) ‐0.0202 (0.3178) ‐0.0177 (0.4242) ‐0.0084 (0.5302) ‐0.0174 (0.6367) ‐0.0065 (0.0023)*** ‐0.0009 (0.0010) ‐0.0063 (0.0078) ‐0.0113 (0.0067)* 0.0044 (0.0298) ‐0.0102 (0.0238)
0.0331 (0.0589) ‐0.0212 (0.0166) ‐0.2030 (0.2722) 0.4079 (0.2163)* ‐0.0470 (0.0745) 0.0153 (0.0670) 0.0031 (0.0719) 0.2833 (0.0362)*** 0.0071 (0.0033)** 0.2175 (0.0372)*** 0.0038 (0.0124) 0.0169 (0.0107) ‐0.0121 (0.0124) 0.2099 (0.1239)* 0.1168 (0.1078) 0.2166 (0.2145) 0.3601 (0.3213) 0.4887 (0.4286) 0.6225 (0.5356) 0.7476 (0.6439) ‐0.0055 (0.0023)** ‐0.0015 (0.0011) ‐0.0121 (0.0079) ‐0.0167 (0.0069)** ‐0.0469 (0.0282)* ‐0.0060 (0.0212)
‐0.0512 (0.0462) ‐0.0056 (0.0111) ‐0.3395 (0.2687) 0.2053 (0.1660) ‐0.0111 (0.0511) 0.0148 (0.0449) 0.0453 (0.0515) 0.2907 (0.0323)*** 0.0037 (0.0030) 0.2326 (0.0343)*** ‐0.0052 (0.0113) 0.0114 (0.0095) 0.0038 (0.0112) 0.2629 (0.1251)** ‐0.0122 (0.1061) ‐0.0332 (0.2123) ‐0.0234 (0.3181) ‐0.0225 (0.4246) ‐0.0140 (0.5308) ‐0.0244 (0.6373) ‐0.0066 (0.0023)*** ‐0.0009 (0.0010) ‐0.0061 (0.0078) ‐0.0112 (0.0067)* 0.0045 (0.0299) ‐0.0091 (0.0238)
% university education in 2000 census x 0.0154 0.0002 0.0146 0.0000 (0.0220) (0.0199) (0.0219) (0.0199) year Log median home value in 2000 census x 0.0071 0.0106 0.0070 0.0103 year (0.0039)* (0.0037)*** (0.0039)* (0.0037)*** Log median hh income in 2000 census x ‐0.0046 ‐0.0034 ‐0.0046 ‐0.0031 year (0.0067) (0.0074) (0.0068) (0.0074) Constant 19.1783 20.0544 19.1640 20.0472 (1.4847)*** (1.7318)*** (1.4685)*** (1.7302)*** Unit of observation is a hospital‐year. Data biannually from 1996 to 2008. Regressions include hospital‐specific fixed effects, differenced out at means. Dependent variable is logged total costs
Table 9: Impact of specific technologies on different types of costs Technology
(1) (2) (3) Total diagnostic radiology costs CDR CDSS Basic EMR
(4)
(5)
CPOE
Adopt in previous 2 year period Adopt at least 2 years earlier Observations # of hospitals R‐squared CONTROLS Log inpatient days
0.0101 0.0034 (0.0087) (0.0090) ‐0.0294 ‐0.0193 (0.0111)*** (0.0119) 11119 9914 2551 2278 0.67 0.68 ‐0.6522 ‐0.1212 (0.2085)*** (0.1244) 0.2182 0.0046 (0.1464) (0.0972) 0.0460 0.0088 (0.0144)*** (0.0070) 0.0053 0.0066 (0.0017)*** (0.0015)*** ‐0.0270 ‐0.0088 (0.0117)** (0.0082) 0.0759 0.0791 (0.0224)*** (0.0216)*** 0.0166 0.0064 (0.0113) (0.0110) 0.0037 0.0081 (0.0120) (0.0122) 0.0234 ‐0.0047 (0.0393) (0.0208) ‐0.0413 ‐0.0382 (0.0175)** (0.0170)** 0.0120 0.0228 (0.0431) (0.0413) ‐0.0068 0.0021
0.0141 (0.0127) ‐0.0105 (0.0166)
Advanced EMR 0.0063 (0.0123) ‐0.0231 (0.0159)
Log outpatient visits Log inpatient days x Log inpatient days Log outpatient visits x Log outpatient visits Log inpatient days x Log outpatient visits Log total hospital beds Independent practice association hospital Management service organization hospital Equity model hospital Foundation hospital Log admissions Births (000s)
0.0001 (0.0099) ‐0.0312 (0.0126)**
(6) (7) (8) (9) Total nursing administration costs CDR CDSS Basic EMR CPOE
0.0238 (0.0142)* 0.0337 (0.0195)* 12139 13516 12139 11802 2800 3101 2800 2839 0.66 0.66 0.66 0.36 ‐0.1856 ‐0.5997 ‐0.6734 ‐0.3817 (0.1444) (0.2254)*** (0.2274)*** (0.2058)* ‐0.0191 0.1042 0.1057 ‐0.0936 (0.1086) (0.1475) (0.1530) (0.1180) 0.0075 0.0372 0.0408 0.0171 (0.0158)** (0.0160)** (0.0124) (0.0084) 0.0041 0.0056 0.0058 0.0005 (0.0034) (0.0018)*** (0.0018)*** (0.0026) ‐0.0012 ‐0.0169 ‐0.0173 0.0095 (0.0098) (0.0119) (0.0123) (0.0096) 0.0241 0.0810 0.0764 0.0771 (0.0245)*** (0.0225)*** (0.0241)*** (0.0354) 0.0096 0.0180 0.0191 ‐0.0067 (0.0124) (0.0097)* (0.0104)* (0.0159) 0.0090 0.0090 0.0150 ‐0.0036 (0.0138) (0.0110) (0.0119) (0.0189) 0.0185 0.0832 ‐0.0100 0.0193 (0.0217) (0.0398) (0.0437) (0.0393)** ‐0.0300 ‐0.0405 ‐0.0330 ‐0.0113 (0.0187) (0.0172)** (0.0182)* (0.0362) 0.0477 0.0215 0.0268 0.0162 (0.0457) (0.0392) (0.0407) (0.0529) 0.0005 ‐0.0044 ‐0.0021 ‐0.0025
(10)
0.0191 (0.0151) 0.0379 (0.0211)*
0.0249 (0.0167) 0.0407 (0.0224)*
0.0137 (0.0202) 0.0309 (0.0277)
Advanced EMR 0.0173 (0.0213) 0.0362 (0.0283)
10598 2555 0.35
8929 2143 0.35
14358 3463 0.35
12918 3129 0.34
‐0.3220 (0.2150) ‐0.0630 (0.1272) 0.0140 (0.0127) ‐0.0006 (0.0028) 0.0082 (0.0102) 0.0693 (0.0359)* ‐0.0053 (0.0162) ‐0.0061 (0.0201) 0.0693 (0.0389)* ‐0.0023 (0.0363) 0.0191 (0.0556) ‐0.0049
‐0.4034 (0.2333)* ‐0.0903 (0.1440) 0.0139 (0.0150) ‐0.0043 (0.0063) 0.0191 (0.0140) 0.0578 (0.0390) ‐0.0076 (0.0175) ‐0.0054 (0.0226) 0.0633 (0.0437) 0.0047 (0.0413) ‐0.0065 (0.0581) ‐0.0152
‐0.1386 (0.1807) ‐0.0010 (0.1137) 0.0076 (0.0108) 0.0006 (0.0023) 0.0000 (0.0093) 0.0699 (0.0421)* ‐0.0209 (0.0147) ‐0.0362 (0.0188)* 0.0603 (0.0381) ‐0.0166 (0.0316) 0.0186 (0.0488) 0.0155
‐0.1261 (0.1905) ‐0.0068 (0.1180) 0.0068 (0.0115) 0.0009 (0.0024) 0.0001 (0.0096) 0.0664 (0.0447) ‐0.0160 (0.0156) ‐0.0255 (0.0196) 0.0865 (0.0381)** ‐0.0095 (0.0338) 0.0159 (0.0499) 0.0148
(0.0134) (0.0135) ‐0.0776 ‐0.1165 (0.1170) (0.1176) 0.1280 0.0771 (0.1943) (0.1909) For‐profit ownership ‐0.0664 ‐0.0609 (0.0463) (0.0523) 0.0392 0.0356 Non‐secular nonprofit ownership (0.0311) (0.0337) Non‐profit church 0.0240 0.0235 ownership (0.0433) (0.0467) Log number of discharges 0.2059 0.1945 Medicare (0.0555)*** (0.0517)*** 0.0093 Log number of discharges 0.0055 (0.0041) (0.0042)** Medicaid Log number of discharges 0.1438 0.1727 total (0.0538)*** (0.0563)*** Residency or Member of 0.0008 ‐0.0060 Council Teaching Hosps (0.0176) (0.0179) ‐0.0031 Vertically integrated with ‐0.0049 (0.0111) (0.0112) doctors Vert. integ. w drs (excl. 0.0024 ‐0.0018 integrated salary model) (0.0129) (0.0133) FT physicians / total 0.0833 0.1198 hospital beds (0.0567) (0.0614)* Year 1998 ‐0.0617 ‐0.0578 (0.1339) (0.1222) Year 2000 ‐0.0815 ‐0.0786 (0.2678) (0.2440) Year 2002 ‐0.0891 ‐0.0850 (0.4011) (0.3657) Year 2004 ‐0.1653 ‐0.1527 (0.5346) (0.4877) Year 2006 ‐0.2229 ‐0.2079 (0.6695) (0.6100) Year 2008 ‐0.3030 ‐0.2937 (0.8052) (0.7323) Full time physicians and dentists (000s) Percent births
(0.0146) ‐0.0847 (0.1558) 0.0399 (0.2150) ‐0.0361 (0.0549) 0.0687 (0.0357)* 0.0600 (0.0500) 0.1785 (0.0513)*** 0.0082 (0.0044)* 0.1595 (0.0585)*** 0.0010 (0.0196) 0.0010 (0.0125) ‐0.0093 (0.0149) 0.1038 (0.0709) ‐0.1523 (0.1337) ‐0.2691 (0.2671) ‐0.3744 (0.4004) ‐0.5438 (0.5339) ‐0.6992 (0.6678) ‐0.8833 (0.8016)
(0.0116) ‐0.2072 (0.0985)** 0.1363 (0.1568) ‐0.0422 (0.0414) 0.0456 (0.0302) 0.0385 (0.0393) 0.1836 (0.0440)*** 0.0065 (0.0038)* 0.1878 (0.0443)*** 0.0039 (0.0161) ‐0.0107 (0.0101) 0.0061 (0.0116) 0.1334 (0.0558)** ‐0.0763 (0.1175) ‐0.1097 (0.2346) ‐0.1299 (0.3515) ‐0.2159 (0.4685) ‐0.2867 (0.5862) ‐0.3836 (0.7046)
(0.0115) ‐0.1312 (0.1185) 0.1519 (0.1627) ‐0.0363 (0.0445) 0.0554 (0.0315)* 0.0448 (0.0408) 0.1995 (0.0510)*** 0.0058 (0.0040) 0.1842 (0.0498)*** 0.0015 (0.0167) ‐0.0138 (0.0106) 0.0098 (0.0124) 0.1050 (0.0604)* ‐0.1316 (0.1290) ‐0.2227 (0.2575) ‐0.2980 (0.3858) ‐0.4422 (0.5141) ‐0.5676 (0.6434) ‐0.7206 (0.7732)
(0.0208) (0.0231) (0.0247) 0.1700 0.0947 0.0458 (0.3061) (0.2665) (0.3629) 0.2186 0.2184 0.2076 (0.2420) (0.2554) (0.2715) ‐0.0099 0.0161 0.0872 (0.0835) (0.0971) (0.1035) 0.0233 0.0455 0.1019 (0.0703) (0.0808) (0.0874) 0.1188 0.1574 0.2148 (0.0802) (0.0914)* (0.0984)** 0.1800 0.1810 0.1624 (0.0387)*** (0.0408)*** (0.0392)*** 0.0028 0.0037 0.0022 (0.0054) (0.0061) (0.0063) 0.1881 0.1931 0.1935 (0.0518)*** (0.0558)*** (0.0581)*** ‐0.0155 ‐0.0219 ‐0.0113 (0.0316) (0.0336) (0.0372) ‐0.0324 ‐0.0347 ‐0.0427 (0.0189)* (0.0199)* (0.0227)* 0.0480 0.0532 0.0516 (0.0240)** (0.0258)** (0.0286)* 0.0030 ‐0.0107 ‐0.0148 (0.1211) (0.1254) (0.1492) 0.0768 0.0087 0.1260 (0.2264) (0.2503) (0.2649) 0.1408 ‐0.0038 0.2458 (0.4526) (0.5003) (0.5295) 0.3111 0.0905 0.4696 (0.7968) (0.6807) (0.7523) 0.4187 0.1226 0.6284 (0.9072) (1.0028) (1.0616) 0.5960 0.2248 0.8447 (1.1342) (1.2538) (1.3267) 0.7360 0.2889 1.0349 (1.3622) (1.5056) (1.5936)
(0.0182) 0.0346 (0.2245) ‐0.0876 (0.2321) 0.0771 (0.0683) 0.0931 (0.0582) 0.1206 (0.0697)* 0.1269 (0.0303)*** 0.0102 (0.0054)* 0.2096 (0.0419)*** 0.0071 (0.0286) ‐0.0057 (0.0172) 0.0158 (0.0214) ‐0.0133 (0.1154) ‐0.1029 (0.1987) ‐0.2156 (0.3971) ‐0.2366 (0.5966) ‐0.3159 (0.7955) ‐0.3259 (0.9944) ‐0.3802 (1.1942)
(0.0189) 0.0876 (0.2879) ‐0.0544 (0.2460) 0.0885 (0.0736) 0.0935 (0.0625) 0.1330 (0.0739)* 0.1324 (0.0315)*** 0.0080 (0.0057) 0.2130 (0.0447)*** 0.0140 (0.0303) ‐0.0065 (0.0185) 0.0214 (0.0234) ‐0.0306 (0.1296) ‐0.1512 (0.2208) ‐0.3179 (0.4412) ‐0.3889 (0.6631) ‐0.5186 (0.8840) ‐0.5873 (1.1047) ‐0.6875 (1.3267)
MSA dummy x year
‐0.0010 0.0003 0.0010 ‐0.0009 ‐0.0012 ‐0.0063 ‐0.0043 ‐0.0038 ‐0.0065 ‐0.0071 (0.0027) (0.0027) (0.0030) (0.0024) (0.0026) (0.0047) (0.0049) (0.0054) (0.0044) (0.0047) Log population in 2000 ‐0.0042 ‐0.0055 ‐0.0061 ‐0.0051 ‐0.0055 0.0027 0.0022 0.0013 0.0015 0.0013 census x year (0.0012)*** (0.0010)*** (0.0012)*** (0.0010)*** (0.0011)*** (0.0017) (0.0018) (0.0019) (0.0016) (0.0017) % Hispanic in 2000 census ‐0.0070 ‐0.0016 0.0003 ‐0.0019 ‐0.0013 ‐0.0044 ‐0.0009 ‐0.0054 0.0044 0.0057 x year (0.0077) (0.0073) (0.0077) (0.0068) (0.0073) (0.0131) (0.0139) (0.0144) (0.0132) (0.0146) ‐0.0221 ‐0.0221 ‐0.0216 ‐0.0179 ‐0.0162 ‐0.0076 0.0025 ‐0.0008 0.0032 0.0026 % Black in 2000 census x year (0.0073)*** (0.0073)*** (0.0080)*** (0.0067)*** (0.0072)** (0.0116) (0.0126) (0.0134) (0.0112) (0.0119) % age 65+ in 2000 census x ‐0.0204 ‐0.0135 ‐0.0080 0.0053 0.0114 ‐0.0867 ‐0.0224 ‐0.0578 ‐0.0387 ‐0.0311 year (0.0283) (0.0292) (0.0312) (0.0264) (0.0278) (0.0546) (0.0540) (0.0574) (0.0466) (0.0499) ‐0.0060 ‐0.0038 ‐0.0177 0.0122 ‐0.0084 ‐0.0271 ‐0.0375 ‐0.0280 0.0012 % age 25‐64 in 2000 ‐0.0174 (0.0230) (0.0235) (0.0251) (0.0208) (0.0213) (0.0410) (0.0431) (0.0443) (0.0362) (0.0390) census x year % university education in ‐0.0168 ‐0.0270 ‐0.0344 ‐0.0025 ‐0.0067 0.0802 0.0960 0.0928 0.0437 0.0407 2000 census x year (0.0248) (0.0251) (0.0277) (0.0219) (0.0241) (0.0480)* (0.0532)* (0.0580) (0.0421) (0.0468) Log median home value in 0.0127 0.0134 0.0135 0.0124 0.0137 0.0000 ‐0.0013 0.0015 0.0032 0.0074 (0.0074) (0.0081) (0.0067) (0.0071) 2000 census x year (0.0041)*** (0.0039)*** (0.0045)*** (0.0036)*** (0.0038)*** (0.0071) Log median hh income in 0.0014 0.0010 0.0061 0.0026 0.0019 ‐0.0021 0.0036 ‐0.0029 0.0059 0.0015 2000 census x year (0.0076) (0.0073) (0.0079) (0.0069) (0.0073) (0.0132) (0.0142) (0.0151) (0.0121) (0.0130) Constant 13.1213 11.4766 11.9272 13.2580 13.4514 11.6707 10.9682 11.7732 9.9951 9.8916 (1.0397)*** (0.9232)*** (1.0187)*** (0.9890)*** (0.9949)*** (1.2025)*** (1.2904)*** (1.3846)*** (1.1350)*** (1.1594)*** Unit of observation is a hospital‐year. Data biannually from 1996 to 2008. Regressions include hospital‐specific fixed effects, differenced out at means.
Appendix Table 1: Leads and lags to get timing of impact (Dependent variable is log total costs) Technology Sample
(1) (2) Basic EMR adoption All firms Bottom 3 quartiles IT intensive counties Will adopt in 4 to 6 years 0.0296 0.0296 (0.0102)*** (0.0122)** Will adopt in 2 to 4 years 0.0326 0.0337 (0.0128)** (0.0161)** Will adopt within 2 years 0.0404 0.0392 (0.0153)*** (0.0191)** Adopt in previous 2 year 0.0541 0.0609 period (0.0184)*** (0.0225)*** Adopt 2 to 4 years earlier 0.0438 0.0601 (0.0216)** (0.0261)** Adopt 4 to 6 years earlier 0.0422 0.0701 (0.0248)* (0.0303)** Adopt at least 6 years earlier 0.0431 0.0699 (0.0299) (0.0368)* Observations 9122 4810 # of hospitals 2178 1161 R‐squared 0.80 0.83 CONTROLS Log inpatient days ‐0.2965 ‐0.4140 (0.0896)*** (0.1088)*** Log outpatient visits ‐0.1283 ‐0.1247 (0.0626)** (0.0800) Log inpatient days x Log 0.0201 0.0217 inpatient days (0.0055)*** (0.0070)*** Log outpatient visits x Log 0.0101 0.0066 outpatient visits (0.0040)** (0.0040)* Log inpatient days x Log ‐0.0040 0.0038 outpatient visits (0.0067) (0.0084) Log total hospital beds 0.0848 0.0825
(3)
(4) (5) Advanced EMR adoption Top quartile All firms Bottom 3 IT intensive quartiles IT counties intensive counties 0.0236 0.0349 0.0393 (0.0164) (0.0094)*** (0.0127)*** 0.0227 0.0222 0.0267 (0.0201) (0.0100)** (0.0139)* 0.0260 0.0376 0.0401 (0.0239) (0.0111)*** (0.0161)** 0.0181 0.0557 0.0597 (0.0298) (0.0124)*** (0.0176)*** 0.0453 0.0612 ‐0.0064 (0.0353) (0.0145)*** (0.0206)*** ‐0.0254 0.0444 0.0703 (0.0403) (0.0165)*** (0.0235)*** ‐0.0340 0.0132 0.0491 (0.0484) (0.0213) (0.0273)* 3968 13193 7244 901 3175 1757 0.79 0.80 0.79 ‐0.3154 ‐0.3952 ‐0.2448 (0.1519)** (0.0796)*** (0.1107)** ‐0.0684 ‐0.1541 ‐0.2008 (0.1062) (0.0619)** (0.1049)* 0.0281 0.0193 0.0100 (0.0090)*** (0.0041)*** (0.0056)* 0.0124 0.0068 0.0084 (0.0015)*** (0.0040)** (0.0059)** ‐0.0155 0.0060 0.0075 (0.0124) (0.0051) (0.0085) 0.0851 0.0911 0.1340
(6) Top quartile IT intensive counties 0.0296 (0.0135)** 0.0188 (0.0141) 0.0339 (0.0152)** 0.0458 (0.0175)*** 0.0292 (0.0200) 0.0167 (0.0232) ‐0.0219 (0.0333) 5534 1276 0.80 ‐0.5733 (0.1208)*** ‐0.0984 (0.0922) 0.0303 (0.0044)*** 0.0056 (0.0015)*** 0.0027 (0.0078) 0.0457
Independent practice association hospital Management service organization hospital Equity model hospital Foundation hospital Log admissions Births (000s) Full time physicians and dentists (000s) Percent births For‐profit ownership Non‐secular nonprofit ownership Non‐profit church ownership Log number of discharges Medicare Log number of discharges Medicaid Log number of discharges total Residency or Member of Council Teaching Hospitals Vertically integrated with doctors Vert. integ. w drs (excl. integrated salary model) FT physicians / total hospital
(0.0180)*** 0.0007
(0.0209)*** ‐0.0028
(0.0275)*** 0.0027
(0.0167)*** 0.0039
(0.0192)*** ‐0.0022
(0.0240)* 0.0115
(0.0078) 0.0180 (0.0079)** ‐0.0035 (0.0188) ‐0.0328 (0.0150)** 0.0132 (0.0264) 0.0034 (0.0084) ‐0.3722 (0.1855)** 0.1142 (0.1158) ‐0.0219 (0.0371) 0.0181 (0.0271) 0.0699 (0.0369)* 0.1364 (0.0443)*** 0.0075 (0.0025)*** 0.2343 (0.0520)*** 0.0099 (0.0120) 0.0099 (0.0082) ‐0.0012 (0.0105) 0.2454
(0.0114) ‐0.0009 (0.0116) ‐0.0290 (0.0291) ‐0.0192 (0.0175) 0.0144 (0.0304) 0.0204 (0.0192) ‐0.7511 (0.4187)* 0.1504 (0.1544) ‐0.0001 (0.0489) 0.0069 (0.0324) 0.0465 (0.0479) 0.2136 (0.0445)*** 0.0059 (0.0033)* 0.1999 (0.0442)*** 0.0024 (0.0140) 0.0075 (0.0112) 0.0066 (0.0129) 0.4247
(0.0111) 0.0322 (0.0104)*** 0.0166 (0.0218) ‐0.0397 (0.0244) ‐0.0170 (0.0526) ‐0.0024 (0.0105) 0.0915 (0.1847) 0.0735 (0.1816) 0.0048 (0.0602) 0.0394 (0.0526) 0.1025 (0.0638) 0.0944 (0.0471)** 0.0070 (0.0035)** 0.2832 (0.0848)*** 0.0150 (0.0151) 0.0135 (0.0117) ‐0.0116 (0.0142) 0.0001
(0.0071) 0.0083 (0.0076) ‐0.0112 (0.0196) ‐0.0323 (0.0140)** ‐0.0008 (0.0244) 0.0060 (0.0074) ‐0.3658 (0.1447)** 0.1789 (0.0968)* ‐0.0023 (0.0286) 0.0339 (0.0215) 0.0824 (0.0298)*** 0.1325 (0.0337)*** 0.0043 (0.0022)* 0.2204 (0.0375)*** 0.0082 (0.0103) 0.0046 (0.0069) 0.0098 (0.0086) 0.2580
(0.0104) ‐0.0023 (0.0114) ‐0.0460 (0.0287) ‐0.0365 (0.0187)* ‐0.0010 (0.0280) 0.0302 (0.0207) ‐0.8041 (0.4535)* 0.1886 (0.1447) 0.0057 (0.0357) 0.0157 (0.0251) 0.0393 (0.0384) 0.1335 (0.0457)*** 0.0029 (0.0031) 0.2306 (0.0446)*** ‐0.0011 (0.0150) 0.0021 (0.0093) 0.0113 (0.0116) 0.4452
(0.0096) 0.0179 (0.0099)* 0.0095 (0.0237) ‐0.0154 (0.0198) ‐0.0032 (0.0451) 0.0038 (0.0087) 0.0428 (0.1432) 0.0216 (0.1349) 0.0277 (0.0475) 0.0686 (0.0416)* 0.1416 (0.0506)*** 0.1318 (0.0491)*** 0.0028 (0.0031) 0.2413 (0.0642)*** 0.0135 (0.0129) 0.0031 (0.0104) 0.0103 (0.0120) 0.0273
beds Year 1998
(0.0815)*** (0.0965)*** (0.0833) (0.0711)*** (0.0998)*** (0.0701) ‐0.0219 ‐0.0967 0.1220 ‐0.1084 ‐0.3182 0.1808 (0.0798) (0.0951) (0.1356) (0.0710) (0.0911)*** (0.1233) Year 2000 ‐0.0491 ‐0.2182 0.2568 ‐0.2216 ‐0.6535 0.3719 (0.1588) (0.1903) (0.2685) (0.1420) (0.1828)*** (0.2455) Year 2002 ‐0.0432 ‐0.2947 0.4186 ‐0.2989 ‐0.9503 0.5944 (0.2377) (0.2851) (0.4019) (0.2127) (0.2741)*** (0.3672) Year 2004 ‐0.0593 ‐0.3960 0.5623 ‐0.4000 ‐1.2706 0.7954 (0.3171) (0.3802) (0.5363) (0.2836) (0.3655)*** (0.4896) 1.0210 Year 2006 ‐0.0478 ‐0.4612 0.7220 ‐0.4742 ‐1.5624 (0.3966) (0.4756) (0.6700) (0.3547) (0.4572)*** (0.6122)* Year 2008 ‐0.0441 ‐0.5337 0.8604 ‐0.5605 ‐1.8651 1.2184 (0.4771) (0.5713) (0.8065) (0.4260) (0.5488)*** (0.7356)* MSA dummy x year ‐0.0052 ‐0.0051 ‐0.0080 ‐0.0059 ‐0.0067 ‐0.0080 (0.0019)*** (0.0022)** (0.0042)* (0.0018)*** (0.0023)*** (0.0034)** Log population in 2000 ‐0.0017 ‐0.0026 ‐0.0009 ‐0.0018 ‐0.0026 ‐0.0011 census x year (0.0007)** (0.0012)** (0.0012) (0.0007)*** (0.0012)** (0.0010) ‐0.0087 ‐0.0025 ‐0.0165 ‐0.0054 ‐0.0034 ‐0.0060 % Hispanic in 2000 census (0.0058) (0.0062) (0.0116) (0.0053) (0.0064) (0.0100) x year % Black in 2000 census ‐0.0147 ‐0.0137 ‐0.0205 ‐0.0093 ‐0.0076 ‐0.0139 x year (0.0049)*** (0.0064)** (0.0074)*** (0.0045)** (0.0059) (0.0069)** % age 65+ in 2000 census x ‐0.0386 ‐0.0263 ‐0.0603 ‐0.0356 ‐0.0302 ‐0.0400 year (0.0219)* (0.0286) (0.0358)* (0.0197)* (0.0267) (0.0308) ‐0.0415 ‐0.0211 ‐0.0158 ‐0.0317 0.0185 % age 25‐64 in 2000 census x ‐0.0402 year (0.0158)** (0.0197)** (0.0275) (0.0150) (0.0208) (0.0247) % university education in ‐0.0061 ‐0.0113 0.0087 ‐0.0065 ‐0.0312 0.0409 2000 census x year (0.0166) (0.0235) (0.0303) (0.0143) (0.0208) (0.0255) Log median home value in 0.0103 0.0150 0.0040 0.0132 0.0207 0.0038 2000 census x year (0.0030)*** (0.0042)*** (0.0047) (0.0026)*** (0.0037)*** (0.0039) 0.0007 ‐0.0009 0.0003 0.0049 ‐0.0076 Log median household income 0.0013 (0.0050) (0.0061) (0.0084) (0.0046) (0.0059) (0.0073) in 2000 census x year Constant 15.3022 15.2933 15.5743 16.0945 15.0773 17.2294 (0.5709)*** (0.6625)*** (1.0392)*** (0.6488)*** (1.0003)*** (0.9580)*** Unit of observation is a hospital‐year. Data biannually from 1996 to 2008. Regressions include hospital‐specific fixed effects, differenced out at means.
Appendix Table 2: No controls Technology
(1) Basic EMR adoption
Adopt in previous 2 year period
0.2981 (0.0079)*** 0.5837 (0.0081)***
Adopt at least 2 years earlier Adopt in previous 2 year period x IT intensive county Adopt at least 2 years earlier x IT intensive county Adopt in previous 2 year period x Programmer 1996 Adopt at least 2 years earlier x Programmer 1996 Observations # of hospitals R‐squared CONTROLS Top quartile IT using county x year
12093 2298 0.37
(2) Advanced EMR adoption 0.3785 (0.0092)*** 0.5941 (0.0114)***
17631 3367 0.15
(3) Basic EMR adoption 0.2979 (0.0106)*** 0.5993 (0.0112)*** ‐0.2784 (0.0157)*** ‐0.6123 (0.0183)***
(4) Advanced EMR adoption 0.3827 (0.0144)*** 0.6156 (0.0179)*** ‐0.3311 (0.0184)*** ‐0.5804 (0.0240)***
11542 2155 0.53
16933 3184 0.38
0.0696 (0.0016)***
0.0659 (0.0012)***
(5) Basic EMR adoption 0.2807 (0.0127)*** 0.5725 (0.0124)*** ‐0.2542 (0.0239)*** ‐0.5773 (0.0368)***
5256 837 0.46
(6) Advanced EMR adoption 0.4055 (0.0145)*** 0.6037 (0.0184)***
(7) Basic EMR adoption
‐0.3778 (0.0254)*** ‐0.5801 (0.0387)***
0.2922 (0.0151)*** 0.5955 (0.0149)*** ‐0.2530 (0.0208)*** ‐0.5603 (0.0248)*** ‐0.0960 (0.0252)*** ‐0.2214 (0.0411)***
(8) Advanced EMR adoption 0.4204 (0.0198)*** 0.6449 (0.0255)*** ‐0.3229 (0.0243)*** ‐0.5585 (0.0324)*** ‐0.1790 (0.0276)*** ‐0.2491 (0.0428)***
7462 1200 0.25
5147 819 0.59
7326 1177 0.45
0.0612 0.0584 (0.0023)*** (0.0016)*** Programmer in 1996 x year 0.0666 0.0666 0.0253 0.0288 (0.0037)*** (0.0025)*** (0.0046)*** (0.0035)*** Constant 17.4994 17.2661 17.4944 17.2990 18.1014 17.8641 18.0514 17.8526 (0.0051)*** (0.0094)*** (0.0056)*** (0.0098)*** (0.0069)*** (0.0126)*** (0.0073)*** (0.0125)*** Unit of observation is a hospital‐year. Data biannually from 1996 to 2008. Regressions include hospital‐specific fixed effects, differenced out at means. Dependent variable is logged total costs
Appendix Table 3: Include missing controls and add a dummy for “missing data” Technology
(1) Basic EMR adoption
Adopt in previous 2 year period
0.0256 (0.0067)*** 0.0035 (0.0084)
Adopt at least 2 years earlier Adopt in previous 2 year period x IT intensive county Adopt at least 2 years earlier x IT intensive county Adopt in previous 2 year period x Programmer 1996 Adopt at least 2 years earlier x Programmer 1996 Observations # of hospitals R‐squared CONTROLS Missing discharge data Missing hospital type data
(2) Advanced EMR adoption 0.0412 (0.0075)*** 0.0275 (0.0092)***
(3) Basic EMR adoption 0.0326 (0.0091)*** 0.0283 (0.0116)** ‐0.0172 (0.0133) ‐0.0568 (0.0168)***
Log inpatient days x Log inpatient days Log outpatient visits x Log outpatient
0.0157 (0.0092)* ‐0.0026 (0.0120)
(6) Advanced EMR adoption 0.0519 (0.0113)*** 0.0403 (0.0149)***
‐0.0006 (0.0200) ‐0.0213 (0.0333)
(7) Basic EMR adoption
‐0.0363 (0.0208)* ‐0.0251 (0.0300)
0.0226 (0.0119)* 0.0241 (0.0165) ‐0.0101 (0.0171) ‐0.0488 (0.0223)** ‐0.0021 (0.0202) ‐0.0172 (0.0330)
(8) Advanced EMR adoption 0.0649 (0.0162)*** 0.0636 (0.0212)*** ‐0.0231 (0.0203) ‐0.0426 (0.0261) ‐0.0330 (0.0212) ‐0.0163 (0.0296)
17329 3322 0.75
11467 2146 0.77
16791 3171 0.75
5199 831 0.80
7356 1189 0.77
5130 819 0.80
7281 1176 0.78
0.8555 (0.0796)*** 0.0014 (0.0093)
0.8846 (0.0651)*** ‐0.0149 (0.0076)*
0.8985 (0.0829)*** 0.0023 (0.0093) 0.0016 (0.0020)
0.9210 (0.0675)*** ‐0.0141 (0.0077)* ‐0.0014 (0.0014)
1.2566 (0.1799)*** ‐0.0170 (0.0125)
1.4966 (0.1679)*** ‐0.0291 (0.0108)***
‐0.4362 (0.0845)*** ‐0.0775 (0.0509) 0.0272 (0.0050)*** 0.0071
0.0028 (0.0035) ‐0.3487 ‐0.3930 (0.0640)*** (0.2590) ‐0.0817 ‐0.1303 (0.0462)* (0.2636) 0.0207 0.0314 (0.0034)*** (0.0159)** 0.0065 0.0169
1.2214 (0.1733)*** ‐0.0143 (0.0125) ‐0.0017 (0.0025) 0.0027 (0.0035) ‐0.3940 (0.2713) ‐0.1317 (0.2494) 0.0330 (0.0161)** 0.0185
1.4707 (0.1659)*** ‐0.0280 (0.0108)*** ‐0.0038 (0.0018)** 0.0024 (0.0024) ‐0.4357 (0.2115)** ‐0.2397 (0.1703) 0.0151 (0.0108) 0.0063
Programmer in 1996 x year
Log outpatient visits
(5) Basic EMR adoption
11909 2269 0.76
Top quartile IT using county x year
Log inpatient days
(4) Advanced EMR adoption 0.0516 (0.0119)*** 0.0484 (0.0141)*** ‐0.0193 (0.0150) ‐0.0389 (0.0184)**
‐0.3839 (0.0719)*** ‐0.0962 (0.0458)** 0.0232 (0.0047)*** 0.0071
‐0.3218 (0.0576)*** ‐0.0912 (0.0427)** 0.0187 (0.0033)*** 0.0066
0.0024 (0.0024) ‐0.4332 (0.2181)** ‐0.2440 (0.1701) 0.0149 (0.0113) 0.0064
visits (0.0020)*** Log inpatient days x Log outpatient visits 0.0001 (0.0056) Log total hospital beds 0.0566 (0.0190)*** Independent practice association 0.0072 hospital (0.0076) Management service organization 0.0175 (0.0085)** hospital Equity model hospital ‐0.0075 (0.0168) Foundation hospital ‐0.0355 (0.0167)** Log admissions 0.1575 (0.0281)*** Births (000s) 0.0074 (0.0090) Full time physicians and dentists (000s) ‐0.1951 (0.2018) Percent births 0.1371 (0.1017) For‐profit ownership ‐0.0456 (0.0311) Non‐secular nonprofit ownership 0.0173 (0.0226) Non‐profit church ownership 0.0342 (0.0319) Log number of discharges Medicare 0.0850 (0.0189)*** 0.0027 Log number of discharges Medicaid (0.0022) Log number of discharges total 0.0180 (0.0165) Residency or Member of Council Teaching 0.0013 Hospitals (0.0120) Vertically integrated with doctors 0.0096 (0.0081)
(0.0013)*** 0.0008 (0.0041) 0.0772 (0.0154)*** 0.0126 (0.0067)* 0.0091 (0.0074) ‐0.0172 (0.0185) ‐0.0242 (0.0152) 0.1685 (0.0226)*** 0.0080 (0.0078) ‐0.1826 (0.1613) 0.1471 (0.0796)* ‐0.0214 (0.0230) 0.0122 (0.0181) 0.0417 (0.0262) 0.0857 (0.0161)*** 0.0033 (0.0021) 0.0210 (0.0141) 0.0121 (0.0113) 0.0059 (0.0069)
(0.0020)*** ‐0.0020 (0.0063) 0.0698 (0.0188)*** 0.0066 (0.0075) 0.0186 (0.0085)** ‐0.0059 (0.0161) ‐0.0353 (0.0169)** 0.1791 (0.0316)*** 0.0033 (0.0090) ‐0.1876 (0.2040) 0.1162 (0.0998) ‐0.0379 (0.0309) 0.0144 (0.0228) 0.0309 (0.0319) 0.0814 (0.0193)*** 0.0019 (0.0022) 0.0263 (0.0168) 0.0044 (0.0118) 0.0108 (0.0082)
(0.0013)*** ‐0.0001 (0.0045) 0.0867 (0.0154)*** 0.0125 (0.0067)* 0.0091 (0.0074) ‐0.0156 (0.0184) ‐0.0236 (0.0154) 0.1807 (0.0243)*** 0.0050 (0.0078) ‐0.1693 (0.1627) 0.1329 (0.0799)* ‐0.0182 (0.0230) 0.0111 (0.0182) 0.0404 (0.0262) 0.0815 (0.0161)*** 0.0028 (0.0021) 0.0289 (0.0142)** 0.0134 (0.0110) 0.0065 (0.0070)
(0.0115) ‐0.0182 (0.0162) 0.0974 (0.0312)*** 0.0124 (0.0098) 0.0215 (0.0103)** ‐0.0185 (0.0257) ‐0.0545 (0.0195)*** 0.2260 (0.0653)*** ‐0.0046 (0.0177) ‐0.0234 (0.3569) 0.2463 (0.2695) ‐0.1305 (0.0514)** ‐0.0138 (0.0351) ‐0.0401 (0.0410) 0.1167 (0.0283)*** 0.0078 (0.0033)** 0.0256 (0.0227) 0.0014 (0.0124) 0.0194 (0.0112)*
(0.0014)*** 0.0151 (0.0144) 0.1005 (0.0265)*** 0.0152 (0.0093) 0.0087 (0.0095) ‐0.0478 (0.0284)* ‐0.0648 (0.0249)*** 0.1905 (0.0493)*** ‐0.0039 (0.0131) ‐0.0851 (0.3472) 0.1373 (0.1900) ‐0.0509 (0.0317) 0.0038 (0.0253) 0.0173 (0.0318) 0.1268 (0.0239)*** 0.0073 (0.0032)** 0.0442 (0.0206)** 0.0061 (0.0136) 0.0093 (0.0103)
(0.0112)* ‐0.0219 (0.0162) 0.1145 (0.0272)*** 0.0149 (0.0097) 0.0207 (0.0103)** ‐0.0166 (0.0246) ‐0.0549 (0.0191)*** 0.2712 (0.0644)*** ‐0.0026 (0.0175) ‐0.0268 (0.3563) 0.0897 (0.2258) ‐0.1202 (0.0506)** ‐0.0213 (0.0359) ‐0.0445 (0.0412) 0.1036 (0.0276)*** 0.0073 (0.0033)** 0.0338 (0.0224) 0.0016 (0.0122) 0.0178 (0.0112)
(0.0014)*** 0.0147 (0.0144) 0.1117 (0.0249)*** 0.0166 (0.0094)* 0.0077 (0.0094) ‐0.0467 (0.0283)* ‐0.0643 (0.0248)*** 0.2109 (0.0490)*** ‐0.0035 (0.0132) ‐0.0828 (0.3486) 0.0592 (0.1801) ‐0.0465 (0.0314) 0.0011 (0.0257) 0.0171 (0.0319) 0.1147 (0.0233)*** 0.0071 (0.0032)** 0.0522 (0.0202)*** 0.0058 (0.0132) 0.0083 (0.0102)
Vert. integ. w drs (excl. integrated salary model) FT physicians / total hospital beds
‐0.0044 (0.0104) 0.2013 (0.0841)** ‐0.1352 (0.0840) ‐0.2775 (0.1675)* ‐0.3910 (0.2508) ‐0.5271 (0.3345) ‐0.6340 (0.4183) ‐0.7713 (0.5018) ‐0.0036 (0.0020)* ‐0.0016 (0.0008)** 0.0001 (0.0055) ‐0.0139 (0.0049)*** ‐0.0252 (0.0214) ‐0.0518 (0.0152)*** ‐0.0015 (0.0171) 0.0099 (0.0031)*** 0.0079 (0.0052) 16.5171 (0.4645)***
0.0061 (0.0082) 0.2062 (0.0684)*** ‐0.0965 (0.0744) ‐0.2007 (0.1481) ‐0.2724 (0.2218) ‐0.3692 (0.2958) ‐0.4410 (0.3699) ‐0.5462 (0.4437) ‐0.0029 (0.0017)* ‐0.0023 (0.0006)*** 0.0022 (0.0050) ‐0.0118 (0.0041)*** ‐0.0235 (0.0185) ‐0.0401 (0.0141)*** 0.0198 (0.0141) 0.0109 (0.0026)*** 0.0044 (0.0046) 15.9968 (0.3881)***
‐0.0058 (0.0105) 0.1960 (0.0850)** ‐0.1200 (0.0830) ‐0.2477 (0.1655) ‐0.3473 (0.2478) ‐0.4681 (0.3306) ‐0.5595 (0.4133) ‐0.6848 (0.4958) ‐0.0038 (0.0020)** ‐0.0015 (0.0008)* ‐0.0028 (0.0058) ‐0.0143 (0.0049)*** ‐0.0275 (0.0215) ‐0.0563 (0.0153)*** 0.0084 (0.0177) 0.0093 (0.0032)*** 0.0080 (0.0052) 16.4314 (0.5395)***
0.0058 (0.0082) 0.1998 (0.0690)*** ‐0.0855 (0.0746) ‐0.1781 (0.1486) ‐0.2389 (0.2226) ‐0.3234 (0.2970) ‐0.3842 (0.3713) ‐0.4816 (0.4454) ‐0.0031 (0.0017)* ‐0.0021 (0.0007)*** ‐0.0021 (0.0052) ‐0.0115 (0.0041)*** ‐0.0274 (0.0186) ‐0.0456 (0.0144)*** 0.0283 (0.0143)** 0.0107 (0.0026)*** 0.0043 (0.0047) 15.9371 (0.4262)***
‐0.0140 (0.0136) 0.1769 (0.1556) ‐0.0175 (0.1113) ‐0.0451 (0.2214) ‐0.0353 (0.3313) ‐0.0399 (0.4421) ‐0.0505 (0.5526) ‐0.0874 (0.6632) ‐0.0041 (0.0025) ‐0.0016 (0.0011) ‐0.0044 (0.0073) ‐0.0143 (0.0069)** ‐0.0525 (0.0249)** ‐0.0082 (0.0183) ‐0.0014 (0.0226) 0.0098 (0.0044)** ‐0.0008 (0.0071) 16.3218 (1.0282)***
‐0.0020 (0.0120) 0.2008 (0.1496) ‐0.0259 (0.1024) ‐0.0533 (0.2045) ‐0.0533 (0.3060) ‐0.0651 (0.4081) ‐0.0871 (0.5105) ‐0.1410 (0.6123) ‐0.0039 (0.0023)* ‐0.0029 (0.0010)*** 0.0077 (0.0075) ‐0.0072 (0.0060) ‐0.0090 (0.0236) ‐0.0200 (0.0209) 0.0225 (0.0204) 0.0097 (0.0038)** 0.0009 (0.0066) 17.1961 (1.6134)***
‐0.0144 (0.0137) 0.1791 (0.1552) ‐0.0197 (0.1105) ‐0.0512 (0.2199) ‐0.0459 (0.3291) ‐0.0540 (0.4391) ‐0.0665 (0.5488) ‐0.1066 (0.6586) ‐0.0043 (0.0025)* ‐0.0009 (0.0011) ‐0.0076 (0.0075) ‐0.0139 (0.0069)** ‐0.0551 (0.0245)** ‐0.0109 (0.0189) 0.0178 (0.0229) 0.0083 (0.0044)* 0.0003 (0.0071) 15.9457 (0.9427)***
‐0.0004 (0.0120) 0.2031 (0.1498) Year 1998 ‐0.0370 (0.1028) Year 2000 ‐0.0751 (0.2053) ‐0.0864 Year 2002 (0.3073) Year 2004 ‐0.1091 (0.4099) Year 2006 ‐0.1412 (0.5128) ‐0.2068 Year 2008 (0.6149) MSA dummy x year ‐0.0041 (0.0023)* Log population in 2000 census x year ‐0.0022 (0.0010)** 0.0026 % Hispanic in 2000 census (0.0076) x year % Black in 2000 census ‐0.0064 x year (0.0060) % age 65+ in 2000 census x year ‐0.0099 (0.0233) ‐0.0255 % age 25‐64 in 2000 census x year (0.0217) % university education in 2000 census x 0.0382 year (0.0205)* Log median home value in 2000 census x 0.0091 (0.0039)** year Log median hh income in 2000 census x 0.0017 year (0.0066) Constant 16.9888 (1.5915)*** Unit of observation is a hospital‐year. Data biannually from 1996 to 2008. Regressions include hospital‐specific fixed effects, differenced out at means. Dependent variable is logged total costs
Appendix Table 4: Missing EMR set to zero. Also includes missing controls and adds a dummy for “missing data” Technology
(1) Basic EMR adoption
Adopt in previous 2 year period
0.0278 (0.0062)*** 0.0088 (0.0072)
Adopt at least 2 years earlier Adopt in previous 2 year period x IT intensive county Adopt at least 2 years earlier x IT intensive county Adopt in previous 2 year period x Programmer 1996 Adopt at least 2 years earlier x Programmer 1996 Observations # of hospitals R‐squared CONTROLS Missing discharge data Missing hospital type data
(2) Advanced EMR adoption 0.0374 (0.0073)*** 0.0200 (0.0090)**
Log inpatient days x Log inpatient days Log outpatient visits x Log outpatient
(5) Basic EMR adoption 0.0140 (0.0090) ‐0.0064 (0.0103)
‐0.0223 (0.0148) ‐0.0449 (0.0181)**
‐0.0034 (0.0177) ‐0.0237 (0.0253)
(6) Advanced EMR adoption 0.0485 (0.0112)*** 0.0338 (0.0147)** ‐0.0399 (0.0196)** ‐0.0319 (0.0278)
(7) Basic EMR adoption 0.0204 (0.0113)* 0.0189 (0.0142) ‐0.0094 (0.0158) ‐0.0459 (0.0187)** ‐0.0023 (0.0179) ‐0.0155 (0.0249)
(8) Advanced EMR adoption 0.0643 (0.0161)*** 0.0641 (0.0213)*** ‐0.0345 (0.0200)* ‐0.0188 (0.0274) ‐0.0284 (0.0200) ‐0.0552 (0.0255)**
22369 4277 0.76
21737 4100 0.76
21737 4100 0.76
9594 1549 0.79
9594 1549 0.79
9495 1532 0.79
9495 1532 0.79
0.8330 (0.0569)*** ‐0.0113 (0.0068)*
0.8321 (0.0569)*** ‐0.0107 (0.0068)
0.8692 (0.0591)*** ‐0.0103 (0.0068) 0.0011 (0.0013) 0.0031 (0.0019) ‐0.3431 (0.0588)*** ‐0.0834 (0.0421)** 0.0205 (0.0031)*** 0.0068
0.8675 (0.0590)*** ‐0.0103 (0.0068) 0.0001 (0.0012) 0.0029 (0.0018) ‐0.3444 (0.0588)*** ‐0.0821 (0.0420)* 0.0206 (0.0031)*** 0.0067
1.4447 (0.1406)*** ‐0.0265 (0.0095)***
1.4392 (0.1399)*** ‐0.0251 (0.0095)***
1.4368 (0.1402)*** ‐0.0250 (0.0095)*** ‐0.0010 (0.0017)
1.4288 (0.1392)*** ‐0.0247 (0.0094)*** ‐0.0018 (0.0016)
0.0027 (0.0020) ‐0.3338 (0.1618)** ‐0.1544 (0.1054) 0.0145 (0.0109) 0.0060
0.0025 (0.0018) ‐0.3364 (0.1613)** ‐0.1495 (0.1049) 0.0147 (0.0109) 0.0059
‐0.3347 (0.1596)** ‐0.1616 (0.1052) 0.0140 (0.0106) 0.0060
‐0.3320 (0.1589)** ‐0.1608 (0.1046) 0.0139 (0.0106) 0.0060
Programmer in 1996 x year
Log outpatient visits
0.0350 (0.0086)*** 0.0335 (0.0099)*** ‐0.0151 (0.0125) ‐0.0511 (0.0143)***
(4) Advanced EMR adoption 0.0494 (0.0119)*** 0.0445 (0.0140)***
22369 4277 0.76
Top quartile IT using county x year
Log inpatient days
(3) Basic EMR adoption
‐0.3299 (0.0528)*** ‐0.0948 (0.0389)** 0.0192 (0.0030)*** 0.0068
‐0.3298 (0.0528)*** ‐0.0932 (0.0390)** 0.0192 (0.0030)*** 0.0067
visits (0.0013)*** Log inpatient days x Log outpatient visits 0.0007 (0.0038) Log total hospital beds 0.0779 (0.0140)*** Independent practice association 0.0087 hospital (0.0060) Management service organization 0.0144 (0.0065)** hospital Equity model hospital ‐0.0136 (0.0149) Foundation hospital ‐0.0341 (0.0137)** Log admissions 0.1737 (0.0201)*** Births (000s) 0.0081 (0.0068) Full time physicians and dentists (000s) ‐0.1631 (0.0939)* Percent births 0.1857 (0.0742)** For‐profit ownership ‐0.0161 (0.0213) Non‐secular nonprofit ownership 0.0199 (0.0159) Non‐profit church ownership 0.0345 (0.0233) Log number of discharges Medicare 0.0782 (0.0140)*** 0.0032 Log number of discharges Medicaid (0.0019) Log number of discharges total 0.0219 (0.0124)* Residency or Member of Council Teaching ‐0.0010 Hospitals (0.0101) Vertically integrated with doctors 0.0031 (0.0060)
(0.0013)*** 0.0006 (0.0038) 0.0772 (0.0140)*** 0.0088 (0.0060) 0.0140 (0.0065)** ‐0.0140 (0.0149) ‐0.0338 (0.0137)** 0.1739 (0.0201)*** 0.0081 (0.0068) ‐0.1620 (0.0938)* 0.1857 (0.0741)** ‐0.0154 (0.0213) 0.0194 (0.0159) 0.0345 (0.0232) 0.0784 (0.0140)*** 0.0032 (0.0019)* 0.0216 (0.0124)* ‐0.0008 (0.0101) 0.0031 (0.0060)
(0.0013)*** ‐0.0006 (0.0041) 0.0865 (0.0139)*** 0.0084 (0.0060) 0.0150 (0.0065)** ‐0.0130 (0.0146) ‐0.0338 (0.0139)** 0.1838 (0.0212)*** 0.0054 (0.0068) ‐0.1550 (0.0943) 0.1824 (0.0746)** ‐0.0153 (0.0212) 0.0178 (0.0158) 0.0316 (0.0232) 0.0747 (0.0140)*** 0.0025 (0.0019) 0.0294 (0.0125)** 0.0004 (0.0100) 0.0037 (0.0060)
(0.0013)*** ‐0.0006 (0.0041) 0.0859 (0.0139)*** 0.0084 (0.0060) 0.0140 (0.0065)** ‐0.0132 (0.0148) ‐0.0339 (0.0138)** 0.1836 (0.0212)*** 0.0051 (0.0068) ‐0.1525 (0.0942) 0.1830 (0.0745)** ‐0.0139 (0.0212) 0.0172 (0.0159) 0.0321 (0.0232) 0.0750 (0.0140)*** 0.0026 (0.0019) 0.0288 (0.0125)** 0.0002 (0.0099) 0.0034 (0.0060)
(0.0012)*** 0.0073 (0.0090) 0.0998 (0.0230)*** 0.0072 (0.0079) 0.0134 (0.0084) ‐0.0302 (0.0218) ‐0.0506 (0.0187)*** 0.2076 (0.0443)*** ‐0.0022 (0.0119) ‐0.0629 (0.2661) 0.1628 (0.1853) ‐0.0537 (0.0300)* 0.0063 (0.0230) ‐0.0024 (0.0296) 0.1078 (0.0210)*** 0.0078 (0.0028)*** 0.0543 (0.0191)*** ‐0.0012 (0.0120) 0.0071 (0.0082)
(0.0012)*** 0.0069 (0.0090) 0.1006 (0.0230)*** 0.0085 (0.0080) 0.0124 (0.0084) ‐0.0312 (0.0218) ‐0.0507 (0.0185)*** 0.2079 (0.0440)*** ‐0.0021 (0.0119) ‐0.0652 (0.2622) 0.1570 (0.1853) ‐0.0530 (0.0298)* 0.0041 (0.0227) ‐0.0039 (0.0293) 0.1084 (0.0210)*** 0.0081 (0.0029)*** 0.0529 (0.0191)*** ‐0.0009 (0.0120) 0.0074 (0.0083)
(0.0013)*** 0.0078 (0.0090) 0.1097 (0.0212)*** 0.0079 (0.0079) 0.0130 (0.0084) ‐0.0301 (0.0214) ‐0.0511 (0.0187)*** 0.2262 (0.0434)*** ‐0.0025 (0.0120) ‐0.0645 (0.2663) 0.1078 (0.1762) ‐0.0496 (0.0299)* 0.0043 (0.0234) ‐0.0041 (0.0298) 0.1000 (0.0204)*** 0.0073 (0.0028)** 0.0608 (0.0187)*** ‐0.0009 (0.0119) 0.0069 (0.0082)
(0.0013)*** 0.0078 (0.0090) 0.1102 (0.0212)*** 0.0090 (0.0081) 0.0116 (0.0084) ‐0.0306 (0.0216) ‐0.0510 (0.0184)*** 0.2247 (0.0431)*** ‐0.0025 (0.0120) ‐0.0651 (0.2628) 0.1020 (0.1761) ‐0.0469 (0.0296) 0.0030 (0.0229) ‐0.0038 (0.0294) 0.1006 (0.0203)*** 0.0077 (0.0028)*** 0.0591 (0.0186)*** ‐0.0006 (0.0119) 0.0067 (0.0082)
Vert. integ. w drs (excl. integrated salary model) FT physicians / total hospital beds
0.0023 (0.0072) 0.1870 (0.0475)*** ‐0.0907 (0.0645) ‐0.1957 (0.1284) ‐0.2650 (0.1922) ‐0.3591 (0.2563) ‐0.4297 (0.3204) ‐0.5306 (0.3845) ‐0.0015 (0.0015) ‐0.0023 (0.0006)*** 0.0008 (0.0046) ‐0.0133 (0.0037)*** ‐0.0267 (0.0165) ‐0.0441 (0.0122)*** 0.0150 (0.0121) 0.0089 (0.0024)*** 0.0069 (0.0041)* 16.1177 (0.3650)***
0.0019 (0.0072) 0.1861 (0.0473)*** ‐0.0905 (0.0644) ‐0.1965 (0.1282) ‐0.2663 (0.1920) ‐0.3616 (0.2560) ‐0.4329 (0.3200) ‐0.5349 (0.3840) ‐0.0015 (0.0015) ‐0.0023 (0.0006)*** 0.0011 (0.0046) ‐0.0131 (0.0037)*** ‐0.0270 (0.0164)* ‐0.0446 (0.0121)*** 0.0140 (0.0121) 0.0088 (0.0024)*** 0.0071 (0.0041)* 16.1056 (0.3655)***
0.0018 (0.0073) 0.1810 (0.0477)*** ‐0.0869 (0.0646) ‐0.1876 (0.1286) ‐0.2531 (0.1925) ‐0.3424 (0.2566) ‐0.4092 (0.3208) ‐0.5089 (0.3850) ‐0.0017 (0.0015) ‐0.0022 (0.0006)*** ‐0.0043 (0.0047) ‐0.0134 (0.0037)*** ‐0.0302 (0.0166)* ‐0.0502 (0.0123)*** 0.0167 (0.0125) 0.0091 (0.0024)*** 0.0069 (0.0041)* 16.0037 (0.4045)***
0.0021 (0.0073) 0.1804 (0.0476)*** ‐0.0835 (0.0646) ‐0.1820 (0.1286) ‐0.2446 (0.1925) ‐0.3318 (0.2567) ‐0.3960 (0.3210) ‐0.4939 (0.3851) ‐0.0017 (0.0015) ‐0.0022 (0.0006)*** ‐0.0040 (0.0047) ‐0.0132 (0.0037)*** ‐0.0316 (0.0165)* ‐0.0504 (0.0123)*** 0.0160 (0.0124) 0.0091 (0.0024)*** 0.0068 (0.0041)* 16.0014 (0.4037)***
‐0.0036 (0.0097) 0.1945 (0.1222) 0.0210 (0.0842) 0.0330 (0.1680) 0.0746 (0.2514) 0.1079 (0.3351) 0.1298 (0.4190) 0.1196 (0.5029) ‐0.0036 (0.0019)* ‐0.0024 (0.0008)*** 0.0035 (0.0066) ‐0.0114 (0.0053)** ‐0.0330 (0.0216) ‐0.0182 (0.0175) 0.0281 (0.0165)* 0.0055 (0.0034) 0.0032 (0.0055) 16.1416 (0.6764)***
‐0.0050 (0.0098) 0.1944 (0.1214) 0.0235 (0.0849) 0.0369 (0.1693) 0.0804 (0.2533) 0.1140 (0.3377) 0.1363 (0.4223) 0.1270 (0.5068) ‐0.0036 (0.0019)* ‐0.0024 (0.0008)*** 0.0034 (0.0066) ‐0.0110 (0.0053)** ‐0.0342 (0.0215) ‐0.0197 (0.0174) 0.0265 (0.0165) 0.0057 (0.0035)* 0.0030 (0.0056) 16.1301 (0.6751)***
‐0.0039 (0.0097) 0.1963 (0.1221) 0.0082 (0.0844) 0.0074 (0.1683) 0.0360 (0.2518) 0.0564 (0.3357) 0.0663 (0.4198) 0.0428 (0.5039) ‐0.0037 (0.0019)* ‐0.0019 (0.0009)** ‐0.0028 (0.0068) ‐0.0115 (0.0053)** ‐0.0359 (0.0214)* ‐0.0242 (0.0181) 0.0344 (0.0168)** 0.0055 (0.0035) 0.0037 (0.0055) 16.0140 (0.6823)***
‐0.0039 (0.0098) 0.1959 (0.1213) Year 1998 0.0094 (0.0853) Year 2000 0.0087 (0.1701) 0.0382 Year 2002 (0.2545) Year 2004 0.0580 (0.3393) Year 2006 0.0671 (0.4243) Year 2008 0.0432 (0.5092) ‐0.0037 MSA dummy x year (0.0019)* Log population in 2000 census x year ‐0.0019 (0.0009)** % Hispanic in 2000 census ‐0.0032 x year (0.0068) ‐0.0110 % Black in 2000 census x year (0.0053)** % age 65+ in 2000 census x year ‐0.0359 (0.0214)* ‐0.0263 % age 25‐64 in 2000 census x year (0.0180) % university education in 2000 census x 0.0333 year (0.0167)** Log median home value in 2000 census x 0.0058 year (0.0035)* 0.0036 Log median hh income in 2000 census x year (0.0055) Constant 16.0029 (0.6778)*** Unit of observation is a hospital‐year. Data biannually from 1996 to 2008. Regressions include hospital‐specific fixed effects, differenced out at means. Dependent variable is logged total costs
Appendix Table 5: Missing EMR set to one. Also includes missing controls and adds a dummy for “missing data” Technology
(1) Basic EMR adoption
Adopt in previous 2 year period
0.0278 (0.0062)*** 0.0088 (0.0072)
Adopt at least 2 years earlier Adopt in previous 2 year period x IT intensive county Adopt at least 2 years earlier x IT intensive county Adopt in previous 2 year period x Programmer 1996 Adopt at least 2 years earlier x Programmer 1996 Observations # of hospitals R‐squared CONTROLS Missing discharge data Missing hospital type data
(2) Advanced EMR adoption 0.0374 (0.0073)*** 0.0200 (0.0090)**
(3) Basic EMR adoption 0.0350 (0.0086)*** 0.0335 (0.0099)*** ‐0.0151 (0.0125) ‐0.0511 (0.0143)***
Log inpatient days x Log inpatient days Log outpatient visits x Log outpatient
0.0140 (0.0090) ‐0.0064 (0.0103)
(6) Advanced EMR adoption 0.0485 (0.0112)*** 0.0338 (0.0147)**
‐0.0034 (0.0177) ‐0.0237 (0.0253)
(7) Basic EMR adoption
‐0.0399 (0.0196)** ‐0.0319 (0.0278)
0.0204 (0.0113)* 0.0189 (0.0142) ‐0.0094 (0.0158) ‐0.0459 (0.0187)** ‐0.0023 (0.0179) ‐0.0155 (0.0249)
(8) Advanced EMR adoption 0.0643 (0.0161)*** 0.0641 (0.0213)*** ‐0.0284 (0.0200) ‐0.0552 (0.0255)** ‐0.0345 (0.0200)* ‐0.0188 (0.0274)
22369 4277 0.76
21737 4100 0.76
21737 4100 0.76
9594 1549 0.79
9594 1549 0.79
9495 1532 0.79
9495 1532 0.79
0.8330 (0.0569)*** ‐0.0113 (0.0068)*
0.8321 (0.0569)*** ‐0.0107 (0.0068)
0.8692 (0.0591)*** ‐0.0103 (0.0068) 0.0011 (0.0013)
0.8675 (0.0590)*** ‐0.0103 (0.0068) 0.0001 (0.0012)
1.4447 (0.1406)*** ‐0.0265 (0.0095)***
1.4392 (0.1399)*** ‐0.0251 (0.0095)***
‐0.3431 (0.0588)*** ‐0.0834 (0.0421)** 0.0205 (0.0031)*** 0.0068
0.0031 (0.0019) ‐0.3444 ‐0.3338 (0.0588)*** (0.1618)** ‐0.0821 ‐0.1544 (0.0420)* (0.1054) 0.0206 0.0145 (0.0031)*** (0.0109) 0.0067 0.0060
1.4368 (0.1402)*** ‐0.0250 (0.0095)*** ‐0.0010 (0.0017) 0.0027 (0.0020) ‐0.3347 (0.1596)** ‐0.1616 (0.1052) 0.0140 (0.0106) 0.0060
1.4288 (0.1392)*** ‐0.0247 (0.0094)*** ‐0.0018 (0.0016) 0.0025 (0.0018) ‐0.3320 (0.1589)** ‐0.1608 (0.1046) 0.0139 (0.0106) 0.0060
Programmer in 1996 x year
Log outpatient visits
(5) Basic EMR adoption
22369 4277 0.76
Top quartile IT using county x year
Log inpatient days
(4) Advanced EMR adoption 0.0494 (0.0119)*** 0.0445 (0.0140)*** ‐0.0223 (0.0148) ‐0.0449 (0.0181)**
‐0.3299 (0.0528)*** ‐0.0948 (0.0389)** 0.0192 (0.0030)*** 0.0068
‐0.3298 (0.0528)*** ‐0.0932 (0.0390)** 0.0192 (0.0030)*** 0.0067
0.0029 (0.0018) ‐0.3364 (0.1613)** ‐0.1495 (0.1049) 0.0147 (0.0109) 0.0059
visits (0.0013)*** Log inpatient days x Log outpatient visits 0.0007 (0.0038) Log total hospital beds 0.0779 (0.0140)*** Independent practice association 0.0087 hospital (0.0060) Management service organization 0.0144 (0.0065)** hospital Equity model hospital ‐0.0136 (0.0149) Foundation hospital ‐0.0341 (0.0137)** Log admissions 0.1737 (0.0201)*** Births (000s) 0.0081 (0.0068) Full time physicians and dentists (000s) ‐0.1631 (0.0939)* Percent births 0.1857 (0.0742)** For‐profit ownership ‐0.0161 (0.0213) Non‐secular nonprofit ownership 0.0199 (0.0159) Non‐profit church ownership 0.0345 (0.0233) Log number of discharges Medicare 0.0782 (0.0140)*** 0.0032 Log number of discharges Medicaid (0.0019) Log number of discharges total 0.0219 (0.0124)* Residency or Member of Council Teaching ‐0.0010 Hospitals (0.0101) Vertically integrated with doctors 0.0031 (0.0060)
(0.0013)*** 0.0006 (0.0038) 0.0772 (0.0140)*** 0.0088 (0.0060) 0.0140 (0.0065)** ‐0.0140 (0.0149) ‐0.0338 (0.0137)** 0.1739 (0.0201)*** 0.0081 (0.0068) ‐0.1620 (0.0938)* 0.1857 (0.0741)** ‐0.0154 (0.0213) 0.0194 (0.0159) 0.0345 (0.0232) 0.0784 (0.0140)*** 0.0032 (0.0019)* 0.0216 (0.0124)* ‐0.0008 (0.0101) 0.0031 (0.0060)
(0.0013)*** ‐0.0006 (0.0041) 0.0865 (0.0139)*** 0.0084 (0.0060) 0.0150 (0.0065)** ‐0.0130 (0.0146) ‐0.0338 (0.0139)** 0.1838 (0.0212)*** 0.0054 (0.0068) ‐0.1550 (0.0943) 0.1824 (0.0746)** ‐0.0153 (0.0212) 0.0178 (0.0158) 0.0316 (0.0232) 0.0747 (0.0140)*** 0.0025 (0.0019) 0.0294 (0.0125)** 0.0004 (0.0100) 0.0037 (0.0060)
(0.0013)*** ‐0.0006 (0.0041) 0.0859 (0.0139)*** 0.0084 (0.0060) 0.0140 (0.0065)** ‐0.0132 (0.0148) ‐0.0339 (0.0138)** 0.1836 (0.0212)*** 0.0051 (0.0068) ‐0.1525 (0.0942) 0.1830 (0.0745)** ‐0.0139 (0.0212) 0.0172 (0.0159) 0.0321 (0.0232) 0.0750 (0.0140)*** 0.0026 (0.0019) 0.0288 (0.0125)** 0.0002 (0.0099) 0.0034 (0.0060)
(0.0012)*** 0.0073 (0.0090) 0.0998 (0.0230)*** 0.0072 (0.0079) 0.0134 (0.0084) ‐0.0302 (0.0218) ‐0.0506 (0.0187)*** 0.2076 (0.0443)*** ‐0.0022 (0.0119) ‐0.0629 (0.2661) 0.1628 (0.1853) ‐0.0537 (0.0300)* 0.0063 (0.0230) ‐0.0024 (0.0296) 0.1078 (0.0210)*** 0.0078 (0.0028)*** 0.0543 (0.0191)*** ‐0.0012 (0.0120) 0.0071 (0.0082)
(0.0012)*** 0.0069 (0.0090) 0.1006 (0.0230)*** 0.0085 (0.0080) 0.0124 (0.0084) ‐0.0312 (0.0218) ‐0.0507 (0.0185)*** 0.2079 (0.0440)*** ‐0.0021 (0.0119) ‐0.0652 (0.2622) 0.1570 (0.1853) ‐0.0530 (0.0298)* 0.0041 (0.0227) ‐0.0039 (0.0293) 0.1084 (0.0210)*** 0.0081 (0.0029)*** 0.0529 (0.0191)*** ‐0.0009 (0.0120) 0.0074 (0.0083)
(0.0013)*** 0.0078 (0.0090) 0.1097 (0.0212)*** 0.0079 (0.0079) 0.0130 (0.0084) ‐0.0301 (0.0214) ‐0.0511 (0.0187)*** 0.2262 (0.0434)*** ‐0.0025 (0.0120) ‐0.0645 (0.2663) 0.1078 (0.1762) ‐0.0496 (0.0299)* 0.0043 (0.0234) ‐0.0041 (0.0298) 0.1000 (0.0204)*** 0.0073 (0.0028)** 0.0608 (0.0187)*** ‐0.0009 (0.0119) 0.0069 (0.0082)
(0.0013)*** 0.0078 (0.0090) 0.1102 (0.0212)*** 0.0090 (0.0081) 0.0116 (0.0084) ‐0.0306 (0.0216) ‐0.0510 (0.0184)*** 0.2247 (0.0431)*** ‐0.0025 (0.0120) ‐0.0651 (0.2628) 0.1020 (0.1761) ‐0.0469 (0.0296) 0.0030 (0.0229) ‐0.0038 (0.0294) 0.1006 (0.0203)*** 0.0077 (0.0028)*** 0.0591 (0.0186)*** ‐0.0006 (0.0119) 0.0067 (0.0082)
Vert. integ. w drs (excl. integrated salary model) FT physicians / total hospital beds
‐0.0039 (0.0098) 0.1959 (0.1213) Year 1998 0.0094 (0.0853) Year 2000 0.0087 (0.1701) 0.0382 Year 2002 (0.2545) Year 2004 0.0580 (0.3393) Year 2006 0.0671 (0.4243) Year 2008 0.0432 (0.5092) ‐0.0037 MSA dummy x year (0.0019)* Log population in 2000 census x year ‐0.0019 (0.0009)** % Hispanic in 2000 census ‐0.0032 x year (0.0068) ‐0.0110 % Black in 2000 census x year (0.0053)** % age 65+ in 2000 census x year ‐0.0359 (0.0214)* ‐0.0263 % age 25‐64 in 2000 census x year (0.0180) % university education in 2000 census x 0.0333 year (0.0167)** Log median home value in 2000 census x 0.0058 year (0.0035)* 0.0036 Log median hh income in 2000 census x year (0.0055) Constant 16.0049 (0.6777)*** Unit of observation is a hospital‐year. Data biannually from 1996 to 2008. Regressions include hospital‐specific fixed effects, differenced out at means. Dependent variable is logged total costs
0.0023 (0.0072) 0.1870 (0.0475)*** ‐0.0907 (0.0645) ‐0.1957 (0.1284) ‐0.2650 (0.1922) ‐0.3591 (0.2563) ‐0.4297 (0.3204) ‐0.5306 (0.3845) ‐0.0015 (0.0015) ‐0.0023 (0.0006)*** 0.0008 (0.0046) ‐0.0133 (0.0037)*** ‐0.0267 (0.0165) ‐0.0441 (0.0122)*** 0.0150 (0.0121) 0.0089 (0.0024)*** 0.0069 (0.0041)* 16.1006 (0.3651)***
0.0019 (0.0072) 0.1861 (0.0473)*** ‐0.0905 (0.0644) ‐0.1965 (0.1282) ‐0.2663 (0.1920) ‐0.3616 (0.2560) ‐0.4329 (0.3200) ‐0.5349 (0.3840) ‐0.0015 (0.0015) ‐0.0023 (0.0006)*** 0.0011 (0.0046) ‐0.0131 (0.0037)*** ‐0.0270 (0.0164)* ‐0.0446 (0.0121)*** 0.0140 (0.0121) 0.0088 (0.0024)*** 0.0071 (0.0041)* 16.0927 (0.3658)***
0.0018 (0.0073) 0.1810 (0.0477)*** ‐0.0869 (0.0646) ‐0.1876 (0.1286) ‐0.2531 (0.1925) ‐0.3424 (0.2566) ‐0.4092 (0.3208) ‐0.5089 (0.3850) ‐0.0017 (0.0015) ‐0.0022 (0.0006)*** ‐0.0043 (0.0047) ‐0.0134 (0.0037)*** ‐0.0302 (0.0166)* ‐0.0502 (0.0123)*** 0.0167 (0.0125) 0.0091 (0.0024)*** 0.0069 (0.0041)* 16.0026 (0.4047)***
0.0021 (0.0073) 0.1804 (0.0476)*** ‐0.0835 (0.0646) ‐0.1820 (0.1286) ‐0.2446 (0.1925) ‐0.3318 (0.2567) ‐0.3960 (0.3210) ‐0.4939 (0.3851) ‐0.0017 (0.0015) ‐0.0022 (0.0006)*** ‐0.0040 (0.0047) ‐0.0132 (0.0037)*** ‐0.0316 (0.0165)* ‐0.0504 (0.0123)*** 0.0160 (0.0124) 0.0091 (0.0024)*** 0.0068 (0.0041)* 15.9953 (0.4039)***
‐0.0036 (0.0097) 0.1945 (0.1222) 0.0210 (0.0842) 0.0330 (0.1680) 0.0746 (0.2514) 0.1079 (0.3351) 0.1298 (0.4190) 0.1196 (0.5029) ‐0.0036 (0.0019)* ‐0.0024 (0.0008)*** 0.0035 (0.0066) ‐0.0114 (0.0053)** ‐0.0330 (0.0216) ‐0.0182 (0.0175) 0.0281 (0.0165)* 0.0055 (0.0034) 0.0032 (0.0055) 16.1506 (0.6760)***
‐0.0050 (0.0098) 0.1944 (0.1214) 0.0235 (0.0849) 0.0369 (0.1693) 0.0804 (0.2533) 0.1140 (0.3377) 0.1363 (0.4223) 0.1270 (0.5068) ‐0.0036 (0.0019)* ‐0.0024 (0.0008)*** 0.0034 (0.0066) ‐0.0110 (0.0053)** ‐0.0342 (0.0215) ‐0.0197 (0.0174) 0.0265 (0.0165) 0.0057 (0.0035)* 0.0030 (0.0056) 16.1276 (0.6750)***
‐0.0039 (0.0097) 0.1963 (0.1221) 0.0082 (0.0844) 0.0074 (0.1683) 0.0360 (0.2518) 0.0564 (0.3357) 0.0663 (0.4198) 0.0428 (0.5039) ‐0.0037 (0.0019)* ‐0.0019 (0.0009)** ‐0.0028 (0.0068) ‐0.0115 (0.0053)** ‐0.0359 (0.0214)* ‐0.0242 (0.0181) 0.0344 (0.0168)** 0.0055 (0.0035) 0.0037 (0.0055) 16.0296 (0.6821)***
Figure 1a: % rise in costs from 2 years earlier, by timing of basic EMR adoption 0.18 0.16 0.14 0.12 0.1 0.08 0.06 0.04 0.02 0 All locations
Non‐IT intensive locations 2 years before adopt
IT‐Intensive locations year adopt
No programmers Has programmers in 1996 in 1996 2 years after
Figure 1b: % rise in costs from 2 years earlier, by timing of advanced EMR adoption 0.18 0.16 0.14 0.12 0.1 0.08 0.06 0.04 0.02 0 All locations
Non‐IT intensive locations 2 years before adopt
IT‐Intensive locations year adopt
No programmers Has programmers in 1996 in 1996 2 years after
Figure 2a: Coefficients by years from adoption 0.12 0.1 0.08 0.06 0.04 0.02 0 ‐0.02 ‐0.04 6 years before adoption
4 years before adoption
2 years before adoption
Adoption period
Basic EMR
2 years after 4 years after 6 or more adoption adoption years after adoption
Advanced EMR
Figure 2b: Coefficients by years from Basic EMR adoption 0.25 0.15 0.05 ‐0.05 ‐0.15
6 years before adoption
4 years before adoption
2 years before adoption
Adoption period
non IT intensive location
2 years after 4 years after 6 or more adoption adoption years after adoption IT intensive location
Figure 2c: Coefficients by years from Advanced EMR adoption 0.15 0.1 0.05 0 ‐0.05 ‐0.1 6 years before adoption
4 years before adoption
2 years before adoption
Adoption period
non IT intensive location
2 years after 4 years after 6 or more adoption adoption years after adoption IT intensive location
Error bars show 95% confidence intervals. Full set of coefficients in Appendix Table 1
