Anesthesia Information Management Systems

Advances in Anesthesia 26 (2008) 121–136

ADVANCES IN ANESTHESIA Anesthesia Information Management Systems Michael M. Vigoda, MD, MBAa,*, David M. Feinstein, MDb,c,d a

Center for Informatics and Perioperative Management, Department of Anesthesiology, Perioperative Medicine, and Pain Management, Miller School of Medicine, University of Miami, 1611 NW 12th Avenue (C-301), Miami, FL 33136, USA b Harvard Medical School, Boston, MA, USA c Information Technology and Simulation Training, Department of Anesthesia, Critical Care, and Pain Medicine d Carl J. Shapiro Simulation and Skills Center, Beth Israel Deaconess Medical Center, Boston, MA, USA

CURRENT STATE OF SYSTEMS History Although less than 10% of institutions have anesthesia information management systems (AIMS), the value of aggregating perioperative data was realized long ago. As early as 1934, Rovenstine [1] reported on the benefits of using surgical and anesthesia data to report clinical outcomes. In the 1970s, several academic centers developed in-house systems [2–4] reflecting early use of computing power in medical record keeping. With the development of mini- and microcomputers in the 1980s, it became possible to use a small computer in an operating room (OR) to collect data automatically. The promise of automating record keeping was strong but frequently outstripped the reality of day-to-day clinical use. As early as 1985, it was said that ‘‘to date no pursuit in anesthesiology technology has claimed more and delivered less than the search of a computerized anesthesia record’’ [5]. Commercialization of these systems (in the 1990s) signaled a new direction. Several companies (eg, Arkive, Compurecord, Picis, Saturn, Aimcare) emerged with the specific intent of automating anesthesia documentation. Although several companies lacked sufficient funding (and ultimately left the market), others became firmly established players in this small but growing medical software segment. Interest increased when the Anesthesia Patient Safety Foundation endorsed and advocated (in 2001) ‘‘the use of automated record keeping in the perioperative period and the subsequent retrieval and analysis of the data to improve patient safety’’ [6]. Currently, there are two distinct groups of vendors—enterprise-wide electronic medical record (EMR) vendors (eg, Cerner, GE) and smaller niche *Corresponding author. E-mail address: [email protected] (M.M. Vigoda). 0737-6146/08/$ – see front matter doi:10.1016/j.aan.2008.07.011

ª 2008 Elsevier Inc. All rights reserved.

122

VIGODA & FEINSTEIN

players, including Docusys, Safersleep, iMDsoft, Philips, Picis, and Draeger. The absence of standards for AIMS would suggest that the current group of vendors is likely to compete with each other for the next 5 to 10 years. Given the low market penetration of AIMS, the push to implement EMRs, and the high switching costs, it seems reasonable to anticipate that even those vendors with a small market share may risk the capital necessary to stay in the game in the hope of future revenues. OPPORTUNITIES AND PROMISE Record Keeping Those unfamiliar with AIMS typically regard them as automated anesthesia record-keeping systems (AARKs), although, in fact, this feature merely represents the most basic aspect of an AIMS. The value-enhancing functions (eg, auditing, decision support) are considered after reviewing the benefits (and concerns) of electronic record keeping. Automated documentation Clearly, one benefit of automated record keeping is the potential to free the anesthesiologist from manually transcribing physiologic and ventilator data. Do we pay less attention if we delegate the recording of real-time data? Weinger and colleagues [7] suggested that ‘‘intraoperative teaching increases the workload of the clinician instructor and may reduce vigilance during anesthesia care,’’ although using a small group of anesthesia residents, Loeb [8] concluded that manual record keeping is not required for vigilance. The lack of accuracy of our current paper-based records, at least with regard to physiologic data, was initially documented more than 2 decades ago [9]. Reich and colleagues [10] demonstrated that handwritten records tend to eliminate extreme values by smoothing data. The tendency to produce ‘‘railroad track’’ vital sign data obscures the known hemodynamic responses to common perioperative conditions. Absence of documented variation in heart rate (HR) and blood pressure (BP), particularly during stimulating episodes, such as intubation or extubation, precludes our ability to characterize percentiles of hypertension (during intubation) or hypotension (after intubation while the surgical field is being prepared). In a more general context, invalid data impair quality control efforts. Although pay-for-performance initiatives offer the potential to improve health care delivery, self-reported inaccurate data limit the potential benefits. How common is underreporting? Two studies (one from Germany and the other from the United States) highlight the magnitude [11,12]. Even when using institutional-specific criteria for deviations in real-time data (eg, hypotension, hypertension, bradycardia, tachycardia), anesthesiologists documented these quality assurance (QA) events less than 15% of the time. User-entered documentation Documentation of user-defined events (eg, anesthesia start/end, intubation, laryngoscopy) using a paper-based record is restricted to a small area of writing

ANESTHESIA INFORMATION MANAGEMENT SYSTEMS

123

space for hours of anesthetic care. Limiting the amount of user-entered documentation discourages adequate description of intraoperative events. Absence of comprehensive documentation may have a variety of consequences, including hindering one’s colleagues in planning subsequent management of future anesthetics, inadequate description for intensive care unit (ICU) providers, and medicolegal ramifications in the event of adverse outcome. Electronic charting offers clinicians several benefits when user-entered documentation is required. Scripts (or case-based default descriptions) provide standardized legible documentation. They can also serve as a visual reminder for specific actions, such as the administration of prophylactic antibiotics, checking eyes, and padding pressure points (Fig. 1). Omissions in user-entered documentation Although the anesthetic record is an important clinical and legal record, omission of relevant user-entered documentation has been problematic with paperbased and electronic records. Despite agreement on which variables should be documented, practicing anesthesiologists using paper-based records commonly omit basic information pertaining to airway (49%), allergies (84%), medications

Fig. 1. Case-based default documentation. Courtesy of Michael O’Reilly, University of Michigan.

124

VIGODA & FEINSTEIN

(69%), and diagnosis (46%) [13]. Recording rates of intraoperative variables are generally better and range from 100% (eg, anesthesiologist’s name, start time of anesthesia) to 24% (eg, estimated blood loss). Studies conducted in a simulated environment using paper records demonstrated that incomplete charting occurred in almost two thirds of records. Incomplete charting was not related to an anesthesiologist’s age, level of training, or number of years in practice. There were, however, marked differences in charting discrepancies of commonly documented physiologic variables (HR, BP, end-tidal carbon dioxide [ETCO2], and pulse oximetry [SpO2]), which were related to level of training [14]. Interestingly, the situation is not remarkably different with electronic records. Driscoll and colleagues [15] found that even basic descriptors of anesthetic care were not always documented. Accurate documentation is most likely related to the ease of using the AIMS software. Entries requiring free text entry (eg, allergies [64%], endotracheal tube depth [59%], electrocardiographic [ECG] rhythm [86%]) were documented less often than those that used a touch screen (eg, mask ventilation [85%], any notation of intravenous access [84%], laryngoscopic grade of view [92%]). Preoperative documentation Electronic records are also helpful for preoperative documentation for several reasons. Beyond the capabilities for remote and simultaneous access, most AIMS are able to copy forward information from previous evaluations, thus minimizing the amount of transcription that would normally be required for patients who frequently undergo anesthesia. In addition, for those facilities in which much of the preoperative history is extracted from nursing or surgical history forms, an EMR can be an efficient method of transferring data that the anesthesiologist currently transcribes from other forms. Although the ability to copy or paste eases one’s work load, there are several caveats, which, although less applicable in an AIMS, are highly relevant to those working in institutions using an enterprise-wide EMR. One study reported that up to one quarter of patient charts have examinations that have been copied in their entirety or in part from previous documentation [16]. Copying or pasting entries can occasionally result in meaningless and verbose (even humorous) documentation [17]. Moreover, some payers do not accept such entries as evidence of care, as illustrated by a Centers for Medicare and Medicaid Services (CMS)–contracted provider’s view of cloned notations: Documentation is considered cloned when each entry in the medical record for a beneficiary is worded exactly like or similar to the previous entries. Cloning also occurs when medical documentation is exactly the same from beneficiary to beneficiary. It would not be expected that every patient had the exact same problem, symptoms, and required the exact same treatment. Cloned documentation does not meet medical necessity requirements for coverage of services rendered because of the lack of specific individual

ANESTHESIA INFORMATION MANAGEMENT SYSTEMS

125

information. All documentation in the medical record must be specific to the patient and his or her situation at the time of the encounter. Cloning of documentation is considered a misrepresentation of the medical necessity requirement for coverage of services. Identification of this type of documentation leads to denial of services for lack of medical necessity and recoupment of all overpayments made [18].

Barriers to Adoption Several barriers to adoption exist. Hospitals, which typically are the purchasers of AIMS, are concerned with the initial start-up costs, security, confidentiality, and need to interconnect (interface) an AIMS with existing systems. Anesthesiologists are concerned about negative impact to work flow, recording of artifact data into the AIMS database, and possibly medicolegal implications. Abouleish and Conlay [19] noted that some of the touted benefits, such as automated charge capture, may give hospitals access to an anesthesiology group’s billing and revenue data. Auditing Auditing electronic records can form the basis for monitoring the delivery of clinical care and process flows within the perioperative environment. Operational Addressing day-of-surgery scheduling problems is a common occurrence in any OR. Which room is going to finish first? Has the procedure in room 3 actually started? By scanning for user-entered events (eg, surgery start, incision, surgery end), the software can automatically generate an electronic ‘‘whiteboard’’ display. By incorporating real-time data, such as physiologic and vital sign data, accurate monitoring of case progression can be determined from the OR control desk or remote locations [20]. Clinical Electronic documentation facilitates monitoring of clinical practice by using regular reports that can be generated automatically. It is possible to monitor compliance with prophylactic antibiotics, intraoperative use of beta-blockers, hypothermia, and redosing of antibiotics protocols. Mining AIMS databases has been used to predict intraoperative events, such as hypotension after induction [21], difficult mask ventilation [22], or the risk for postoperative nausea and vomiting (PONV) [23]. Timely feedback When data are in an electronic format, querying the AIMS database can lead to new approaches to monitoring clinical practice—principally by using timely feedback. Automating the analysis of AIMS data and providing timely feedback to anesthesiologists has been demonstrated to be an effective method of changing physician behavior. Vigoda and Lubarskky [24] reported that feedback within 24 hours was effective in decreasing preattested documentation from 75% to 0.5% within 6 months. O’Reilly and colleagues [25] provided

126

VIGODA & FEINSTEIN

specific feedback to individual providers with the goal of ensuring that patients received prophylactic antibiotics in a timely manner. Biweekly e-mails, based on reports generated from AIMS data, were sent to all providers. Each provider received an email summarizing the entire group’s performance in addition to his or her individual results. Within 1 year, compliance with antibiotic guidelines steadily increased from 69% to 92%. Monitoring of documentation necessary for billing is also possible and may be especially helpful for those practices with anesthesia care teams. At the authors’ institution, all records from the prior 24 hours are scanned for the necessary CMS documentation (eg, chart reviewed and anesthetic plan discussed, present for critical events). Records that are missing one or more required elements are flagged, and e-mails are sent to individual attendings so that the necessary corrections are made. This retrospective approach required less than 1 week of the programmer’s time. The front office billing clerk’s work load has been reduced by 75%, principally on the basis of not having to review records that contain all the necessary CMS documentation. Real-time notification of missing documentation entries is also possible. There are several advantages in reminding anesthesiologists of documentation deficiencies while the patient is in the OR, including user satisfaction, obviating the need to reopen a record the following day, and a sense that the system is aiding the clinician as opposed to merely monitoring his or her behavior. Real-time feedback can yield impressive results. Sandberg and colleagues [26] demonstrated a significant improvement in the quality of intraoperative documentation in only 2 days. Kheterpal and colleagues [27] reviewed records at specific times during anesthesia (eg, surgical incision, anesthesia end) to determine the presence of valid arterial line pressure values. In the absence of a corresponding procedure note, a member of the care team received a message sent by means of e-mail and to his or her alphanumeric text pager. After a 2-month study, documentation compliance increased from a baseline of 80% to 98%. Extrapolating these results, this single intervention was projected to yield an annual incremental gain of $40,000. Spring and colleagues [28] estimated that automated real-time error detection of required CMS documentation entries increased their annual revenue by $400,000. The most significant improvements noted were a reduction in the percentage of unbillable records from 1.31% to 0.04% and a decrease in the median time to correct documentation from 33 days to 3 days. Decision Support Information-intensive industries (eg, financial, aviation) have used software to assist decision makers, because decisions involve analyzing vast amounts of data or because decisions have to be made in a short period of time. This software is broadly referred to as decision support, and in health care, the most widespread example is computerized physician order entry (CPOE). Although there is tremendous potential for minimizing medication errors and facilitating prescription writing, researchers have reported unanticipated consequences (eg,

ANESTHESIA INFORMATION MANAGEMENT SYSTEMS

127

more or new work for clinicians, unfavorable work flow issues, continuing system demands) after implementation [29,30]. Before incorporating decision support into AIMS (by vendors or by individual institutions), these findings should be kept in mind. Notwithstanding, there are numerous scenarios in which decision support would assist anesthesiologists passively (eg, user initiates action, which retrieves needed information) or actively (eg, information is presented to the user without specifically asking for it). Passive Many institutions already use passive decision support in the form of easily accessible reference materials. These include protocols for advanced cardiac life support or malignant hyperthermia, medications to be given during transplant procedures, and prophylactic antibiotic guidelines and doses. A slightly more active form of decision support would include automated retrieval of prior anesthetic records or a drug dose calculator with weight imported from AIMS application. Other examples include automatic generation of letter templates that can be given to patients in the event of a difficult airway, possible drug reaction, or dental trauma (personal communication, Richard Epstein, 2007). Active Preoperative decision support can assist in the preanesthesia process by identifying specific herbal medications that patients are taking, determining if a patient qualifies for a clinical study, automating the American Heart Association/American College of Cardiology (AHA/ACC) algorithm for evaluation of patients with cardiac risk factors, or as described previously, identification of patients at risk for PONV [24]. Intraoperative decision support could assist anesthesiologists who are supervising multiple rooms so that they are automatically notified by personal digital assistant (PDA) or text pager about changes in a patient’s condition. Other possible applications include identification of possible episodes of light anesthesia or awareness, automated calling for help (eg, code), or predicting wake-up time based on PK analysis of propofol infusion history. After surgery, some measure of quality control could be achieved by automated creation of individualized postanesthesia orders based on standardized templates for particular types of procedures. By accounting for body mass index (BMI) and ideal body weight, in addition to intraoperative medication administration, decision support can provide recommendations for appropriate postoperative orders. This could be especially helpful when used for children (eg, orders of magnitude difference in weight) and the elderly (eg, sensitivity to medications). IMPLEMENTATION Defining the Project The decision to acquire an AIMS has consequences beyond the practice of anesthesia—collaboration between the anesthesia department, medical facility (center), and information services (IS) is essential. Proper planning dictates

128

VIGODA & FEINSTEIN

that system configuration incorporate the immediate and future needs of all the parties involved. From the outset, the project scope must be well defined. Consideration should be given to where, when, and how the AIMS is going to be used (eg, preoperative evaluations, intraoperative charting, postoperative documentation). These decisions depend on the institution’s current implementation and future plans for an EMR. A new system that does not integrate with the other hospital systems may initially be easier to design for the department’s needs but may not meet future institutional needs. What is the rationale for acquiring an AIMS? Some may simply desire a legible anesthetic record. Others look toward creating a robust integrated database that can be queried for departmental and hospital-wide quality improvement and patient safety initiatives. Two questions are relevant: What data are going to be collected and by whom? What reports are going to be generated from these data and for whom? The answers provide a template for understanding the needs that may be met by the AIMS. The common dilemma with EMRs is that although everyone wants the data (for a variety of reasons), no one wants to input the data. Identifying the best system for your institution may be challenging and may best start with an in-depth analysis of available AIMS. The selection process can be quite perplexing—not unlike buying one’s first car or house. From the outset, great care must be taken to understand what the AIMS can do in the institution. Despite vendor demonstrations, promises, and visits to AIMS working sites (with multiple conversations with current AIMS users), the level of understanding acquired for the final selection may still be less than desired. In addition, the negotiation process between the vendor and department or hospital to create a statement of work (SOW) and purchase agreement can involve many months of discussion. Appropriate effort placed during this early process can certainly pay off in the end when the AIMS implementation is under way. The rest of this article assumes that an AIMS has been identified and concentrates on implementation. The Project Team A successful AIMS implementation requires a dedicated collaboration involving the anesthesia department, the hospital or institution, and the AIMS vendor. Each entity should have committed representatives on the AIMS project team. At a minimum, an anesthesia project leader, vendor project leader, and IS project leader need to be identified. This team should be responsible for leading the project through its various phases, including design, configuration, testing, training, rollout, and support. Each of these phases is discussed. The anesthesia project leadership typically comprises an individual (or individuals) in the anesthesia department with designated responsibility for the AIMS project. Along with the vendor and IS AIMS project leadership, they participate in fiscal and clinical decisions along an agreed project time line. The anesthesia project leadership should do their best to understand the

ANESTHESIA INFORMATION MANAGEMENT SYSTEMS

129

AIMS-related needs in their department. This needs assessment should be done with input from department representatives and is most likely handled in AIMS-related departmental committees. The vendor’s representative (vendor project leader) provides the management skills to adhere to project time lines. He or she also has access to a team of vendor experts to bring knowledge and insight into the AIMS, specifically advising how the AIMS can be customized to meet the institution’s needs. Most anesthesia departments do not have experience in putting in an AIMS, and thus rely on expertise from the vendor and hospital IS to help with the implementation. The hospital IS AIMS project leader, if available, can play a key role in the AIMS project. Bridging the gap between vendor and clinicians, the IS team adds the institutional-specific system and computing expertise necessary to integrate hospital IS systems with AIMS and other systems. Their experience, a result of implementing other institutional systems, should be invaluable in the design, testing, teaching, and rollout phases of the project. One other IS member is required—a system administrator who is responsible for the day-to-day upkeep of the AIMS (eg, backup of data, ongoing configuration, database queries, report generation, hardware configuration, training, AIMS-related communication to the users and institution). This individual should have excellent computing, database administration, and communication skills to be the primary resource for the AIMS. Identifying a system administrator early in the process enables that person to acquire the expertise necessary to administer the system. Anesthesia Information Management System Configuration Most AIMS vendors provide guidelines for system configuration. These provide a time line that depends on the level of complexity, integration with the hospital system, and size and scope of the project. Milestones written into the SOW are linked with payments to the vendor. These bilateral contractual discussions are designed to incorporate expectations from the vendor and the institution. The AIMS project team should ensure that these time lines are reasonable and agreed on by all parties so as to move forward with the implementation process. Software configuration Customizing a vendor’s system to fit your institutional needs is no small task. It is commonly believed that the AIMS should mimic the work flow processes used in a paper-based system. Such thinking is usually the result of our instinctive aversion to change. Notwithstanding, the implementation of an AIMS can serve as a time to re-evaluate current work flow processes. It is possible to streamline many aspects of clinical work flow documentation when systems are integrated through collaborative efforts of institutional departments. Most users accept a new system as long as they can understand and appreciate how it is going to have a positive impact on their practice. The practicing anesthesiologist does not want to (or cannot) type documentation entries—and

130

VIGODA & FEINSTEIN

he or she should not have to. Long lists of event descriptors typically result in absent, poor, or incomplete documentation. Organizing medication and event documentation is a time-consuming process but is well worth the effort. Analyzing the user’s work flow (and arranging icons and windows) should be done with the intent of ‘‘Making It Easy to Do It Right’’ [31]. Change is never easy, and strong leadership is required for a smooth transition. Identifying colleagues who are advocates for this change can help with the effort. These individuals may act as ‘‘superusers,’’ who know more than the average user about the workings of the system and can be additional resources for training and configuration design. These resources should be a welcome addition to the AIMS team structure outlined previously. Configuration decisions should ideally be made with input from everyone involved in the process. As much as possible, the team should consider the configurations already in use at other sites to determine if they might satisfy your institution’s needs. The challenge in optimizing a software configuration is in creating intuitive user interfaces (eg, logical navigation using obvious menu choices) that permit user-entered documentation to be completed in a user’s natural work flow. One significant data entry concern is the tradeoff between the use of predetermined text versus free text entry (or typing) by the user. Although free text can certainly add valuable information regarding the provision of care, it should be minimized because it lessens the ability to search the database. Data representation, in addition to the use of those data in the clinical workflow, merits attention. A standardized consistent output is important in paper records and equally so for the EMR. The institution’s medical records forms committee determines the standards, which may limit the range of configuration options. Allowing individuals to customize the anesthesia record output may impede acceptance of this ‘‘new’’ form or jeopardize the process. Regulatory agencies, including the Joint Commission on Accreditation of Healthcare Organizations (JCAHO), generally prohibit abbreviations of medications in order entry. Although this concern is much lessened when using an EMR, it is also a good patient safety practice to develop a list of unacceptable abbreviations. Organizations like the Institute for Safe Medication Practices have developed such lists [32], which may ease this process. Accurate documentation is useful in optimizing hospital and professional fee reimbursement processes [33]. Compliance with CMS requirements can be increased using real-time notification [29] or automated review after the case has been completed [25]. Data may be collected for pay for performance initiatives, such as those developed by the American College of Surgeons National Surgical Quality Improvement Program (NSQIP) or the Surgical Care Improvement Project (SCIP). Simplifying the documentation of clinical processes can thus ultimately reap financial rewards. Hardware configuration Hardware configuration is equally important. A full perioperative AIMS consists of networked application and database servers, printers, and workstations

ANESTHESIA INFORMATION MANAGEMENT SYSTEMS

131

in anesthetizing locations in addition to postanesthesia care units (PACUs), ICUs, and holding and preoperative assessment locations. The hospital IS team should determine the best configuration of database servers that provides for adequate redundancy, minimal failover, security, proper recovery schemes, and proper network connectivity (eg, wired, wireless). Be aware that for some anesthetizing locations, it may be a challenge to implement acceptable ergonomic placement of hardware components for end users of the system. AIMS devices, such as central processing units (CPUs), displays, touch screens, keyboards, mice, and other input devices, are subjected to the harsh OR environment with electrosurgical units (ESUs), blood, and other fluids. Power supplies for the components should be surge protected and isolated for line monitoring. Although some vendors may offer or require their own proprietary or specific hardware, others leave the decision up to the purchaser. Hardware mounting varies depending on physical location and placement of existing equipment. Considerable time may be spent in developing a consensus regarding hardware placement (eg, left side or right side of the anesthesia machine) and the use of an input device (eg, mouse versus trackball versus touchpad). The need for specific cables or wireless access requires assessment of each unique anesthetizing location. Anesthesia Information Management System Testing On completion of system configuration, thorough system testing is required. This time-consuming and laborious process is essential to ensure successful deployment. Again, the hospital IS team can be helpful in determining how best to accomplish this task. Developing simulated scripts and using all facets of the software are important parts of the process. Testing should ensure that correct data are entered into the system and that the data are accurately stored and can subsequently be accurately retrieved or queried. Most often, the emphasis is placed on the software customization and user interface. Although most of the configuration time is spent on software, it is important to remember that the system hardware and network should also be rigorously tested. The hardware testing should reproduce the clinical environment so that other equipment, such as ESUs, anesthesia machines, and monitoring diagnostic medical equipment are all present and represent actual conditions. Software Software testing should be done with users having varying computer and clinical skills to ensure that the AIMS meets the needs of the most users. In real clinical practice, the AIMS user would be multitasking, prioritizing care of the patient over documentation of such care. For this reason, the user interface should be designed to allow quick and easy access to existing data and provide easy ability to enter data into the system. AIMS testing can be done in actual or simulated anesthetizing locations. Feedback from end users testing the system can help to gain perspective on the optimization of the software interface.

132

VIGODA & FEINSTEIN

Hardware In general, the AIMS hardware is configured to receive and record data from hospital systems, physiologic monitors, anesthesia machines, and infusion and other devices, in addition to clinicians. Hardware configurations may vary throughout the AIMS deployment because they depend on the setup of different anesthetizing locations and other areas in which AIMS use occurs. Simulation of power outages and surges and of network and clinical device disconnects should all be part of the testing process. Training Training schemes for AIMS are as varied as the systems themselves. The configuration should be well tested and finalized so that the training and associated materials reflect the final state of the system configuration. Although the approaches may differ, the common goal is to have all users well trained and comfortable with the use of the system before the ‘‘go-live’’ date. The AIMS vendor (and your IS department) can provide guidance for this important phase of the implementation. Typically, the vendor assists with the early training, which includes ‘‘train the trainer’’ programs. Individuals identified as superusers are then able to lead some aspects of the training. Timing and scheduling Ideally, once individuals are trained and proficient, they should be able to continue practicing with the system so that they remain comfortable when the go-live date arrives. Continued proficiency depends on the complexity of the system, the number of trainees, and the timing between training and golive. Adequate training time should be set aside well in advance to minimize disruption to department work flow. Training site and materials If possible, a classroom should be established for training department members in the use of the AIMS. AIMS hardware may be used before deployment on a test system. Some centers may be able to use a simulation environment for testing and training of the AIMS. In some cases, generic training materials may be provided by the vendor. These materials usually need modification based on institutional-specific policies and practices. Materials may include trifold pamphlets, slide presentations, and Web or computer-based training modules. Trainees AIMS training is needed for most people using the system. Depending on the scope of the project, billing personnel, QA staff, other department administrative staff, and hospital IS personnel may also need training on the system. Rollout AIMS rollout strategies are influenced by many factors, including support resources, hardware and network preparedness timing, clinical documentation, and billing requirements. To ease the transition, in-depth planning must be

ANESTHESIA INFORMATION MANAGEMENT SYSTEMS

133

in place to handle all contingencies for hardware, software, network and other failures. IS and AIMS administrative support should be available in abundance during the initial go-live period. The time interval between training and go-live for the clinicians may affect the process. Paper (or original system) backup should always be available to handle unanticipated problems. To account for some of these issues, institutions have successfully implemented a staggered rollout of the AIMS, starting in a clinical or physical area one day and progressing to include other areas over the ensuing rollout time. If the AIMS includes preoperative, intraoperative, and postoperative documentation, the system may be rolled out in coordination with the flow of information so that preoperative documentation would precede intraoperative documentation, for example. Such rollouts seem to be more controlled, and there are usually enough resources to support the transitioning clinicians. If too prolonged, however, the duality of the hybrid system may be confusing to department members and the work flow inconsistencies may be difficult to handle. In addition, vendor support and payment may depend on the rollout as a milestone, and their resources may be limited for prolonged implementation processes. No matter how the rollout happens, the new AIMS should be monitored for accuracy and completeness of the new documentation. The documentation cycle from preoperative care to intraoperative care and postoperative care, QA, and billing should be reviewed. Interfaces with the hospital systems (admission/discharge/ transfer [ADT] feed, pharmacy, laboratories), physiologic monitors, anesthesia machines, and other interfaced devices should all be analyzed to ensure that they are working properly. Support Support for the AIMS varies from one installation to another and depends on the initial system setup. Most vendors have contractual agreements for support, which may include updates, upgrades and day-to-day technical availability. In addition to the AIMS system administrator, department- or hospital-based biomedical engineering, IS, and other designated personnel may be needed to keep the system running and in top form after implementation. If the department has worked independently without involving the IS department, it must rely on the vendor and internal expertise to administer and support the AIMS. Systems conceived and implemented with hospital support may continue to have that support with minimal additional vendor support. Decisions regarding ongoing support should be addressed before moving ahead with the AIMS project. Early planning and discussion should lead to a clearer understanding of costs and responsibilities. Even when successfully implemented, an AIMS is in a dynamic state, allowing for configuration changes that require adaptation to changing clinical, research, and administrative needs. MEDICOLEGAL CONSIDERATIONS Do electronic records help or hurt in defending against claims of malpractice? With less than 10% of institutions using AIMS, there is not a definitive answer.

134

VIGODA & FEINSTEIN

Feldman’s [34] 2004 survey of institutions using AIMS indicated that most believed that AIMS were valuable as a risk management tool. There were 41 cases in which a claim was made when an AIMS was used. In most these, the automated record was helpful in encouraging the plaintiff to drop the claim, facilitating settlement, or helping to document the absence of negligence. In no instance did the automated record hamper the defense, and 90% of respondents would choose to implement EMRs if they were to assume a position at an institution that used paper records. One should note that so-called ‘‘early adopters’’ of any software system are highly motivated to dedicate the resources necessary to ensure successful implementation. There are 27 cases in the American Society of Anesthesiologists Closed Claims Database in which a claim was made in a case in which an AIMS was used. In 4 of these cases, data from the AIMS would have been damaging to the defense. These episodes included the following: 1. SpO2 less than 60% for 1.5 hours and the record showed it accurately 2. Unrecognized esophageal intubation, wherein absence of ETCO2 was accurately recorded for 30 minutes 3. An anesthesiologist’s testimony that was discredited when the automated record indicated that preoxygenation was not performed, contradicting the claim by the doctor that it was 4. Demonstration that the unsuccessful resuscitation of a patient with postpartum hemorrhage was attributable to inadequate care [35]

E-Discovery In a paper-based medical environment, it is clear what we consider the ‘‘anesthetic record’’ to be. In a recent survey [36], however, there was tremendous variation in how often data were stored (range: 15 seconds to 5 minutes), how long a record is kept open (range: until patient leaves the PACU to 2 days after completion of anesthetic), whether users are permitted to edit data that are automatically collected from monitors (one third do not allow it), and whether a person giving a lunch or dinner break is required by departmental policy to add his or her name to the record. These differences, the result of different institutional and vendor policies, may be relevant (or even problematic) in the future. Although some of us may, unfortunately, be familiar with the discovery course of a malpractice claim, transitioning to electronic records should substantially affect this process. Recent amendments to the Federal Rules of Civil Procedure, which govern how civil (and thus malpractice) suits are handled, have had a profound impact on the use of electronic communication and records as evidence. Although an in-depth review of these changes is beyond the scope of this article, all anesthesiologists are encouraged to contact their medical records and risk management departments to understand the e-discovery process better. SUMMARY Adoption of AIMS is increasing as governmental, regulatory, and institutional forces push for EMRs. Anesthesiologists can benefit by proactively identifying

ANESTHESIA INFORMATION MANAGEMENT SYSTEMS

135

features that they consider desirable, participating in institutional selection processes, and contributing to system configuration. The advantages of automated record keeping suggest that once an AIMS is used, few users have an interest in returning to paper. The evolution of these systems should not only help with clinical documentation but ultimately assist clinicians in providing care. Familiarity with how these systems capture, store, and retrieve data is essential in understanding the medicolegal significance of EMRs. Acknowledgments The authors thank Luis I. Rodriguez, Frank Gencorelli, and Sophie Leonard for their editorial assistance. References [1] Rovenstine EE. A method of combining surgical and anesthetic records for statistical purposes. Anesth Analg 1934;13:122–8. [2] deVos CB, Abel MD, Abenstein JP. An evaluation of an automated anesthesia record keeping system. Biomed Sci Instrum 1991;27:219–25. [3] Karliczek GF, de Geus AF, Wiersma G, et al. A computer system for automatic documentation in anesthesia. Int J Clin Monit Comput 1987;4(4):211–21. [4] Osswald PM, Winter D, Hartung HJ, et al. NAPROS: a semiautomatic user-friendly anaesthetic record system. Int J Clin Monit Comput 1987;4(4):231–6. [5] Lees DE. Computerized anesthesia records may have drawbacks: to the editor. Anesthesiology 1985;63(2):236–7. [6] APSF Newsletter 2001. Available at: http://www.apsf.org/resource_center/newsletter/ 2001/winter/. Accessed January 18, 2008. [7] Weinger MB, Reddy SB, Slagle JM. Multiple measures of anesthesia workload during teaching and nonteaching cases. Anesth Analg 2004;98:1419–25. [8] Loeb RG. Manual record keeping is not necessary for anesthesia vigilance. J Clin Monit 1995;11:9–13. [9] Cook R, McDonald JS, Nunziata E. Differences between handwritten and automatic blood pressure records. Anesthesiology 1989;71:385–90. [10] Reich DL, Wood RK, Mattar R, et al. Arterial blood pressure and heart rate discrepancies between handwritten and computerized anesthesia record. Anesth Analg 2000;91(3): 612–6. [11] Benson M, Junger A, Fuchs C, et al. Using an anesthesia information management system to prove a deficit in voluntary reporting of adverse events in a quality assurance program. J Clin Monit Comput 2000;16(3):211–7. [12] Sanborn KV, Castro J, Kuroda M, et al. Detection of intraoperative incidents by electronic scanning of computerized anesthesia records. Comparison with voluntary reporting. Anesthesiology 1996;85:977–87. [13] Tessler MJ, Tsiodras A, Kardash KJ, et al. Documentation on the anesthetic record: correlation with clinically important variables. Can J Anaesth 2006;53(11):1086–91. [14] Devitt JH, Rapanos T, Kurrek M, et al. The anesthetic record: accuracy and completeness. Can J Anaesth 1999;46:122–8. [15] Driscoll WD, Columbia MA, Peterfreund RA. An observational study of anesthesia record completeness using an anesthesia information management system. Anesth Analg 2007;104(6):1454–61. [16] Thielke S, Hammond K, Helbig S. Copying and pasting of examinations within the electronic medical record. Int J Med Inf 2007;76(Suppl 1):122–8. [17] Hirschtick RE. Copy-and-paste. JAMA 2006;295(20):2335–6. [18] Medicare B Update, vol. 3, no. 4, p. 4. First Coast Service Options, Inc. (A CMS Contracted Intermediary and Carrier). Proceedings.

136

VIGODA & FEINSTEIN

[19] Abouleish AE, Conlay L. Automated anesthesia charge capture and submission: wave of the future, or bridge to nowhere? Anesthesiology 2006;105:5–7. [20] Epstein RH, Dexter F, Krol M, et al. Automated correction of room location errors in anesthesia information management systems. Anesth Analg 2008;107:965–71. [21] Reich DL, Hossain S, Krol M, et al. Predictors of hypotension after induction of general anesthesia. Anesth Analg 2005;101:622–8. [22] Kheterpal S, Han R, Tremper KK, et al. Incidence and predictors of difficult and impossible mask ventilation. Anesthesiology 2006;105:885–91. [23] Junger A, Hartmann B, Benson M, et al. The use of an anesthesia information management system for prediction of antiemetic rescue treatment at the postanesthesia care unit. Anesth Analg 2001;92:1203–9. [24] Vigoda MM, Lubarsky DA. The medicolegal importance of enhancing timeliness of documentation when using an anesthesia information system and the response to automated feedback in an academic practice. Anesth Analg 2006;103:131–6. [25] O’Reilly M, Talsma A, VanRiper S. An anesthesia information system designed to provide physician-specific feedback improves timely administration of prophylactic antibiotics. Anesth Analg 2006;103:908–12. [26] Sandberg WS, Sandberg EH, Seim A, et al. Real-time checking of electronic anesthesia records for documentation errors and automatically text messaging clinicians improves quality of documentation. Anesth Analg 2008;106:192–201. [27] Kheterpal S, Gupta R, Bloom JM, et al. Electronic reminders improve procedure documentation compliance and professional fee reimbursement. Anesth Analg 2006;104(3): 592–7. [28] Spring SF, Sandberg WS, Anupama S, et al. Automated documentation error detection and notification improves anesthesia billing performance. Anesthesiology 2006;106(1): 157–63. [29] Koppel R, Metlay JP, Cohen A, et al. Role of computerized physician order entry systems in facilitating medication errors. JAMA 2005;293(10):1197–203. [30] Ash JS, Sittig DF, Poon EG, et al. The extent and importance of unintended consequences related to computerized provider order entry. J Am Med Inform Assoc 2007;14:415–23. [31] James BC. Making it easy to do it right. NEJM 2001;345(3):991–3. [32] Special issue: do not use these dangerous abbreviations or dose designations Available at: http://www.ismp.org/Newsletters/acutecare/articles/20030220_2.asp. Accessed January 8, 2008. [33] Reich DL, Kahn RA, Wax D, et al. Development of a module for point-of-care charge capture and submission using an anesthesia information management system. Anesthesiology 2006;105:179–86. [34] Feldman JM. Do anesthesia information systems (AIS) increase malpractice exposure? Results of a survey. Anesth Analg 2004;99:840–3. [35] Plagenhoef JS in Karin Bierstein’s report. Anesthesia information systems. . .where awareness is good! ASA Newsl 2007;71(3):37–9. [36] Epstein RH, Vigoda MM, Feinstein DM. Anesthesia information management systems: a survey of current implementation policies and practices. Anesth Analg 2007;105:405–11.