Health information technology in US emergency departments
© Springer-Verlag London Ltd 2010
Received: 7 December 2009
Accepted: 10 February 2010
Published: 1 April 2010
Information technology may improve patient safety, and is a focus of health care reform. A minority of emergency departments (EDs) in Massachusetts, and in academic EDs throughout the US, have electronic health records.
Assess health information technology adoption in a nationwide sample of EDs.
We surveyed 69 US EDs, asking site investigators about the availability of health information technology in 2005–2006. Using multiple linear regression, we compared adoption of technology by ED type (emergency medicine residency affiliation, annual census, US region) to assess generalizability of the findings.
Sixty-eight EDs (99%) provided information about health information technology; 75% were affiliated with an emergency medicine residency, and all were urban. Most respondents had applications that simply relay information from one place to another, including patient tracking (74%); ordering tests (laboratory 57%, others 62%); and displaying prior visit notes (79%), ECGs (92%), laboratory (97%), and radiology (99%) results. A minority had more-advanced applications, which seek to modify human behavior, including medication ordering (38%), allergy warnings (19%), and medication cross-reaction warnings (13%), and a few used bar coding (20%). There were no significant differences in technology adoption by ED type.
This and prior studies suggest that some applications—particularly those relevant to modifying clinician behavior—are not widespread in US EDs, while others are. The reasons for this are unknown, but might include expense and unintended consequences. The fact that the emergency medicine community has not rushed to adopt certain applications presents challenges and opportunities.
Health information technology has the potential to improve health care. In early 2009, the American Reinvestment and Recovery Act allocated $19 billion for health information technology funding, and Medicare and Medicaid offer incentives for “meaningful use” of electronic medical records . However, cost, difficulty of implementation, and other barriers impede adoption of such systems, and studies have documented low rates of technology acquisition and implementation in emergency department (ED) and other settings .
A study of residency-affiliated EDs in 2000 found low rates of adoption, with only 7% reporting fully implemented technology for medication error checking, 18% for computerized medication order entry, and 21% for clinical documentation. A 2006 survey of health information technology in all Massachusetts EDs found similar results, with 11% of respondents reporting fully implemented technology for medication error checking . Another national study showed that, during 2001–2003, only 31% of US EDs used electronic medical records in any form .
Our hypothesis is that fewer than half of US EDs had adopted electronic health records systems and related decision-support tools by 2005–2006.
We report on data collected as part of the National ED Safety Study (NEDSS). Details of the study design and data collection have been published previously . NEDSS was coordinated by the Emergency Medicine Network (EMNet, www.emnet-usa.org). We recruited EDs mainly by inviting network members. Since most members are affiliated with an emergency medicine residency program (i.e., are academic EDs), we also recruited non-academic and other EDs not affiliated with EMNet through postings on emergency medicine list servers, by contacting sites directly, and through presentations at emergency medicine meetings. We excluded federal hospitals, children’s hospitals, and hospitals in US territories.
Characteristics of 68 emergency departments surveyeda
Median computerization scoreb
Emergency medicine residency affiliation
Annual census (visit per year)
Our main outcome measure is descriptive: how many EDs have adopted which applications?
Our secondary objective was to determine whether ED characteristics predicted adoption of health information technology. We created two methods for addressing this question.
Firstly, we selected one application that seemed to represent the potential of information technology to improve health care. There is substantial debate about what constitutes “meaningful use” of information technology, and we felt that selecting one crucial application for analysis would be the best way to compare across EDs. For the purpose of this analysis, we selected computerized medication ordering as the application. We chose this application because medication ordering represents a crucial site of potential intervention to improve adherence to recommended care and avoid error. We studied this as an outcome variable by creating a logistic regression model with computerized medication order entry as the dependent variable. We then used the model to determine whether ED characteristics predicted its adoption. We categorized EDs according to emergency medicine residency affiliation (binary), annual census (continuous), and region (categorical: Northeast, Midwest, South, and West). We used multiple logistic regression to model these characteristics as predictors of adoption of computerized medication ordering. We used the Hosmer-Lemeshow goodness of fit test to evaluate for non-linearity.
Secondly, we did not wish to limit our investigation to any one application or combination of applications, as was done in the analysis described above. Thus, we also compared EDs by the raw number of applications they had adopted. We accomplished this by creating a computerization score, as was done in a prior study . This score is simply a count of applications, with one point assigned for each acquired application. We then used multiple linear regression to predict computerization score by the above ED characteristics and planned to consider any characteristic to be a “significant” predictor of the computerization score if the p-value of its beta coefficient was ≤ 0.05. We used the SAS spec option to assess for heteroscedasticity and inspected residual plots for qualitative evidence of a relationship.
Of 69 EDs surveyed, 68 (99%) in 23 US states completed the survey. Table 1 shows the characteristics of EDs surveyed. The participants were generally large urban academic EDs, with 44% located in the Northeast.
Information technology in 68 US emergency departments
Information technology application
Number (%) of EDs reporting availability of each application
Computerized physician medication order entry
Clinical decision support for medication allergies
Clinical decision support for drug-drug interactions
Prior ED visit notes
Laboratory test ordering
Test ordering other than laboratory tests
Bar coding for patient, medication, or sample identification
Our analysis of ED characteristics did not reveal any significant variation in availability of electronic medication ordering or computerization score, by emergency medicine residency affiliation, annual census, or region.
Information technology is ubiquitous in our society. Businesses from restaurants to banks and airlines rely on such technology to process information that has much in common with medical information. Such technology has been recommended strongly for health care settings . Why, then, would our nationwide sample of EDs reveal that only 19% have the capability to check a medication order to see if the patient has a documented allergy to that medication?
Reasons might include apprehension. In the enthusiasm to promulgate the spread of health information technology, the downsides are often under-emphasized . Moreover, purchase and implementation of such technology is not necessarily cost-effective .
The present report echoes the findings of the three prior studies of health information technology adoption in US EDs—namely, slow uptake of certain applications [2, 3]. Only two of these prior studies assessed the adoption of individual applications, such as computerized physician order entry. One was a survey of US emergency medicine residency-affiliated EDs in 2000, and the second was a survey of all Massachusetts EDs in 2006 . Our ability to conduct direct quantitative comparisons across time is limited, because different EDs participated in the surveys. However, certain qualitative observations seem valid. All three studies revealed that a minority of EDs had computerized medication ordering: 38% in the present national study, 15% in the 2006 MA study, and 18% in the 2000 national study. Quantitatively, the increase from 18% to 38% might indicate wider dissemination of these technologies, or might be due to sampling variation. Qualitatively, it seems reasonable to conclude from these three studies that most US EDs do not have electronic medication ordering, or its companion technology, medication error checking. However, this conclusion must be caveated by the fact that large academic EDs were over-represented in these samples.
Studies of health information technology adoption in physicians’ offices and hospitals throughout the country also revealed slow uptake by 2008 [8, 9]. These studies found that only 7.6% of US hospitals have a basic electronic records system, and 1.5% a comprehensive one. (These studies defined “basic” as including only demographic information, CPOE, laboratory and imaging results, and “comprehensive” as including the above, plus clinical notes, plus electronic prescribing, radiographic image display, and decision support.) Computerized provider order entry for medications has been implemented in only 17% of hospitals. Only 4% of physicians’ offices have a fully functional electronic-records system, and 13% have a basic system.
ED-based studies also show that some health information technology applications are widespread, including patient tracking; ordering tests; and displaying prior visit notes, ECGs, and laboratory and radiology results [2, 3, 5]. An analysis of the types of health information technology may improve our understanding of these results. One type of application is computationally simple, merely entering and extracting information from a database, and displaying it. The other type of application requires all of the above processes, plus algorithmic processing that seeks to supplement human cognition with computation. For lack of better terms, we might refer to these two types of application as “passive” and “active” applications, respectively. From a workflow standpoint, passive applications simply convert traditional paper processes into an electronic format. Such technologies include patient tracking; ordering tests; and displaying prior visit notes, ECGs, laboratory and radiology results. Active applications add functions designed to improve outcomes by reducing error or increasing adherence to the standard of care—they seek to change human behavior. These applications often embody fundamental changes in clinical processes, and implementation is intertwined with changes in workflow [6, 10]. Examples include computerized provider order entry and related error-checking functions, and decision support systems. In light of this dichotomization, we can see that the present and prior studies reveal a high rate of adoption of passive applications in US EDs and a low rate of adoption of active applications [2, 3]. (We included laboratory test ordering in the “passive” category for historical reasons. Such systems have been present in most EDs for many years  and were not originally subject to behavior-modification technology.)
An optimistic interpretation of our data would be to celebrate the fact that most US EDs did not invest in expensive, cumbersome, and even potentially risky systems prematurely. Now that a large number of commercial and home-grown systems are available, we should take the time to study them and reach consensus on which would be best for particular environments. We should pay particular attention to the interaction of information systems with human workflows [6, 10]. The Certification Commission for Healthcare Information Technology formed an ED Work Group in 2007, which is in the process of certifying various health information technology platforms for ED use, in concert with the Health Level Seven project . This type of certification process may be an important first step in improving the types of systems that are available, or at least our understanding of what is available.
Our study is limited by the convenience sample of 68 EDs across the US. Participating sites were generally large, urban, academic EDs. Because the underlying patient safety study focused on adult illnesses, the study excluded children’s hospitals. We did assess for variability in technology adoption by ED characteristics, and found no significant variation. This improves our confidence in the generalizability of our data to other EDs in the US, though, again, large academic EDs were over-represented in the sample. Perhaps more saliently, the present results are similar to those of prior studies, from academic EDs throughout the US, and all EDs in Massachusetts . It remains true that we have little data on health information technology adoption in the 1/3 of US EDs with an annual visit volume <10,000. From the perspective of the individual patient, small EDs have the same clinical goals as large EDs. But from a systems perspective, the marginal cost of health information technology in low-volume settings will be high relative to benefit, both financially and in terms of workflow modification and associated risks .
Another limitation is the fact that our assessment was part of a larger survey, and thus, we could not conduct as detailed an inquiry as would have been desirable. In particular, we did not ascertain adoption of health information technology for medication order checking or decision support. However, the low rate of adoption of medication order entry technology allows us to conclude with confidence that these technologies were not available in the majority of respondent EDs.
Active applications involve intertwining of technology and human workflows, and are intended to change behavior. Passive applications involve simple data processing and display. Together with the findings of other studies, our results demonstrate that most US EDs have adopted passive applications but not active applications that have been recommended forcefully [2, 3]. We view this as an opportunity, because careful analysis of available platforms and the pitfalls in their implementation may result in better outcomes and less difficulty in the implementation process. The ideal forum for such analysis remains unclear, though the Certification Commission for Healthcare Information Technology ED work group is a notable effort .
- Blumenthal D (2009) Stimulating the adoption of health information technology. N Engl J Med 360(15):1477–1479PubMedView ArticleGoogle Scholar
- Pallin DJ, Sullivan AF, Auerbach BS, Camargo CA, Jr. Adoption of Information Technology in Massachusetts Emergency Departments. J Emerg Med. 2009 Feb 18 [Epub ahead of print.] 2009Google Scholar
- Burt CW, Hing E. Use of computerized clinical support systems in medical settings: United States, 2001-03. Adv Data. Mar 2 2005(353):1–8Google Scholar
- Sullivan AF, Camargo CA Jr, Cleary PD et al (2007) The national emergency department safety study: study rationale and design. Acad Emerg Med 14(12):1182–1189PubMedView ArticleGoogle Scholar
- Pallin D, Lahman M, Baumlin K (2003) Information technology in emergency medicine residency-affiliated emergency departments. Acad Emerg Med 10(8):848–852PubMedGoogle Scholar
- Ash JS, Sittig DF, Campbell EM, Guappone KP, Dykstra RH. Some unintended consequences of clinical decision support systems. AMIA Annu Symp Proc. 2007:26–30Google Scholar
- Himmelstein DU, Wright A, Woolhandler S. Hospital Computing and the Costs and Quality of Care: A National Study. Am J Med. Nov 16 2009Google Scholar
- DesRoches CM, Campbell EG, Rao SR et al (2008) Electronic health records in ambulatory care—a national survey of physicians. N Engl J Med 359(1):50–60PubMedView ArticleGoogle Scholar
- Jha AK, DesRoches CM, Campbell EG et al (2009) Use of electronic health records in US hospitals. N Engl J Med 360(16):1628–1638PubMedView ArticleGoogle Scholar
- Han YY, Carcillo JA, Venkataraman ST et al (2005) Unexpected increased mortality after implementation of a commercially sold computerized physician order entry system. Pediatrics 116(6):1506–1512PubMedView ArticleGoogle Scholar
- http://www.cchit.org/emergency/ Accessed 3/9/09