- Email: [email protected]
Clinical and practical requirements of online software for anesthesia documentation—an experience report
International Journal of Medical Informatics 57 (2000) 155 – 164
www.elsevier.com/locate/ijmedinf
Clinical and practical requirements of online software for anesthesia documentation—an experience report Matthias Benson a,*, Axel Junger b, Lorenzo Quinzio a, Carsten Fuchs a, Gregor Sciuk a, Achim Michel b, Kurt Marquardt b, Gunter Hempelmann a a
Department of Anesthesiology and Intensi6e Care Medicine, Justus-Liebig Uni6ersity, Rudolf-Buchheim Strasse, 7, 35392 Giessen, Germany b Department of Medical and Administrati6e Data Processing, Justus-Liebig Uni6ersity, Klinikstraße 23, 35392 Giessen, Germany Received 18 May 1999; received in revised form 2 March 2000; accepted 28 March 2000
Abstract The aim of this paper is the presentation of a new version of the anesthesia documentation software, NarkoData, that has been used in routine clinical work in our department as part of an anesthesia information management system (AIMS) since 1995. The performance of this software is presented along with requirements for future development of such a system. The originally used version, NarkoData 3.0, is an online anesthesia documentation software established by the software company ProLogic GmbH. It was primarily developed as a disk-based system for the MacOS operating system (Apple Computer Inc.). Based on our routine experience with the system, a catalogue of requirements was developed that concentrated on improvement in the sequence of work, administration and data management. In 1996, the concepts developed in our department, in close co-operation with medical personnel and the software company, led to a considerable enlargement of the program functions and the subsequent release of a new version of NarkoData. Since 1997, more than 20 000 anesthesia procedures have been recorded annually with this new version at 115 decentralized work stations at our university hospital. © 2000 Elsevier Science Ireland Ltd. All rights reserved. Keywords: Anesthesiology; Automatic anesthesia record; Personal computer; Documentation; Quality assurance
1. Introduction As early as 1895, a formal documentation of an anesthesia procedure was made [1]. * Corresponding author. Tel.: + 49-641-9944494; fax: +49641-9944499. E-mail address: [email protected] (M. Benson).
Since then, anesthesia documentation has been developed continuously and is now an integral part of anesthesiology. At present, the documentation is mainly done on paper. Computer entry of this data is supported by machine-readable forms and/ or done manually after the anesthesia. These methods are very time-consuming [2] and do
1386-5056/00/$ - see front matter © 2000 Elsevier Science Ireland Ltd. All rights reserved. PII: S 1 3 8 6 - 5 0 5 6 ( 0 0 ) 0 0 0 5 9 - 9
156
M. Benson et al. / International Journal of Medical Informatics 57 (2000) 155–164
not meet increasing quality requirements [3– 7]. Another, presently less-used method of documentation, is automated online recording during anesthesia, including automatic data transfer from patient monitoring systems and connection to the hospital information system (HIS). For effective use, an anesthesia documentation software should provide user-friendly, ergonomic interfaces for data entry and extensive evaluation of data. The criteria of the German society for Anesthesiology and Intensive Care Medicine (DGAI) [5,8,9] must be met. Also, safety against failure and the administration of the system during the hospital routine working hours must be ensured. In 1994, the Department of Anesthesiology and Intensive Care Medicine at the JustusLiebig University, Giessen, decided to implement an anesthesia information management system (AIMS). The requirements for such documentation software were put down in a developer’s document. At that time, no adequate commercial product was available to fulfil the high performance requirements, so we signed a co-operation contract with a software company for the development of the program for clinical routine use. Now, after 4 years of experience with the system, the time has come for a critical review and evaluation of the extent of our attaining the objectives.
2. The last version The version NarkoData 3.0 of the anesthesia documentation software, installed in 1995, had initially been developed by the Department of Anesthesiology, Intensive Care Medicine and Analgesia at the University of Bochum (Klinik Bergmannsheil, Head of Department: Professor Dr M. Zenz) in co-operation with the company ProLogic GmbH
(Erkrath, Germany) [10]. The original program recorded medical data during anesthesia and ran as a local application at the workstation. The software was disk-based and designed for the operating system MacOS (Apple Computer, Inc., Cupertino CA) [11]. Completed protocol files were imported into a proprietary database (OMNIS Software, San Bruno CA). Due to the simple database structure, it offered only limited query options. Recording of patient data was carried out with: free text entries multiple-choice fields choice lists and a standard query for the core data pool defined by the DGAI [5]. Many parameters (ventilation, diagnosis) could only be documented once, although they could change over time during the period of anesthesia. Duration of anesthesia procedures was recorded only with the help of very few plausibility checks. Specific data (department information, medication, staff etc.) were read at the start of a new record from a preferences file, which was created by a special administration tool (SuperNarko). The program was not designed to run as part of a network, e.g. it was impossible to import data from the hospital information system (HIS). Administrative effort for the determination of standards, data evaluation and user service in a decentralized hospital was considerably high.
3. Requirements After a thorough check of the options provided by the initial version of the system, a set of requirements was set up for further development. Of particular value was the ex-
M. Benson et al. / International Journal of Medical Informatics 57 (2000) 155–164
Table 1 Main requirements for an online anesthesia documentation software, developed by the Department for Anesthesiology and Intensive Care Medicine of the Justus-Liebig University, Giessen (1) Adaptation to the anesthesiological process Large medical data fields and structure of data input after the work process Permanently visible graphic user interface with time axis (worksheet) Ergonomic data input via mouse, keyboard and shortcuts (hot keys) Representation of the time-dependent parameters on the worksheet with the possibility of multiple documentation Representation of the time-independent parameters in dialogue windows with overview display Data fields for a preoperative and a postoperative visit Sufficient space for comments on the individual parameters Use of the online data transfer from medical equipment (patient monitoring systems, analysis tools) Data import from all available computer systems (HIS, subsystems of other departments, anesthesiological data base) via a standard interface (e.g. HL-7) Help functions with a completeness check available anytime No additional time needed Fast program performance and an adequate process adapted screen presentation (2) Program administration Fixed data structure with configurable medical contents Consequent sequence and logical cheeks of the anesthesia times as guideline for complete documentation Completeness controls for the obligatory entry fields and logical algorithms to check the entered data Password protection of the anesthesiologist in charge High data safety through periodical safety copies High reliability through local program application Combination of network and stand-alone architecture Small administrative effort for installation Small administrative effort for the maintenance of the programs base data Compatible transfer to Windows NT and vice versa
157
(3) Anesthesiological database Use of a relational standard database Clear data structure, division into base tables and process tables Complete and consistent data sets Access to and print of previous anesthesia records at the workstation Export of data from previous anesthesia records into current records Availability of all data for evaluation with a standard SQL tool
perience of daily administration and data management gained from the existing patient data management system (PDMS) Emtek 2000 (Siemens AG, Munich, Germany) at our operative intensive care unit (ICU). In a brief abstract of the 150 page requirement document, the main demands are highlighted: The three most important aspects were an improved adaptation to the course of the anesthetic procedure, simplification of the program administration and improvement of the database structure. Individual points of these three aspects are shown in Table 1. Since then, this new version, NarkoData 4, has been further developed following the requirements and the experience from using NarkoData version 3.0. The new program version was improved continuously. Suggestions from the users were rated and realized whenever possible.
4. The actual version
4.1. Adaptation to anesthesia working procedures The main function expansions of Version 4 are listed in Table 2. For more flexibility, this version is available for MacOS and MS-Windows based PCs. The worksheet with its graphic user interface (Fig. 1) corresponds to the conventional
Fig. 1. The worksheet is the central graphic user interface for displaying and recording time-dependant data. The time-dependent data windows provide a quick overview.
158 M. Benson et al. / International Journal of Medical Informatics 57 (2000) 155–164
M. Benson et al. / International Journal of Medical Informatics 57 (2000) 155–164
Table 2 Decisive functional expansions of the new program version 4 (1) Adaptation to the anesthesiological process Inclusion of previously missing data fields for medical issues Adjusting of time recording, adaptation to the work process Improvement of the graphic user interface ergonomics Import option for data from previous anesthesia records into the current record Expansion of the software drivers for automatic data transfer from additional patient monitoring systems Import option for HIS data via ASCII text, Apple-Events and DDE (Windows) Redesign and enhancement of the printout Improvement of the help functions and overview windows (2) Program administration Improvement of the administration and consistency of the program’s base data Expansion of the network functions of the file management New programming of a compatible Windows NT 6ersion (3) Anesthesiological database Reorganization and data transfer of the data from a hierarchical to a relational standard database Improvement of data consistency through implementation of logical checking algorithms during data input Improvement of data quality through import mechanisms from other subsystems Clear display and printout of previous anesthesia records
paper report. It offers a permanently visible documentation of the complete procedure including the recovery room. Differences in color guarantee clear recognition. The ergonomic and intuitive entry of data during the procedure, e.g. vital data, drugs given and anesthesia times, is directly performed by mouse click. Additional time-dependent parameters can be entered or
159
presented at the desired time in dialogue windows opened by simple double-click. The sequence of data entry is adapted to the actual course of the anesthetic procedure. Pre-configured sequences of dialogue windows, plenty of pre-configured choice lists and logical checks via cross references support a fast and complete documentation. Hence, the required time for staff training is considerably reduced (approximately one working day). The system records time-independent data fields (administrative data, preoperative diagnoses, ICD and ICPM keys, measures and biometrical data) and time-dependent parameters such as vital and respiratory values, drugs given, laboratory data, adverse events, etc. The data fields of the core data set of the DGAI [5] were originally developed for paper documentation and could now be adjusted to accommodate the possibilities of digital online recording. The coding of diagnoses and surgery data follows common coding keys (e.g. ICD, ICPM) stored in the database. An overview window allows an instant orientation in the contents of the dialogue windows at any time (Fig. 1). This is especially helpful for patient transfer. Call of pop-up windows or data entry fields via hot keys or mouse is executed immediately; server waiting times do not affect the use of the system. The program offers the possibility to record preoperative data with a laptop computer at the patient’s bedside in the ward. This pre-operative visit can be printed after an automatic integrity check is carried out, thus providing the ward’s staff with written instructions. This information is made available to the anesthesiologist on the day of the procedure. A similar documentation for the postoperative visit has not yet been realized. If pre-configured, standardized anesthesia record forms are used, the administrative,
160
M. Benson et al. / International Journal of Medical Informatics 57 (2000) 155–164
preoperative and surgery-related data can be entered in 5–10 min. The importation of patient core data and laboratory data from previous anesthesia records cuts the time to 3 – 5 min. The content of the pre-configured anesthesia record forms is usually determined by the doctor in charge.
4.2. System architecture The program itself runs as a local application. Every 2 minutes, two independent safety copies of the anesthesia protocol are recorded in plain text format, one on the local harddisk and the other on another data storage medium. Because the safety copies are made consecutively on two different data storage media, the maximum loss of data is cut to one storage interval (2 min). If the program is running on a network, the second data storage medium is set by default to a server providing all connected clients with record files. Computers, which are not connected to the network, use a floppy disk as second data storage medium. These disks also ensure data exchange between the individual work stations, in case of network or server failure, by loading the anesthesia protocol stored on the floppy disk and automatically saving it on the server when the computer is reconnected to the network. Thus, a mixed architecture of networked and independent computers is possible. Within the computer network, it is possible to import data from the HIS (via AppleEvents or DDE), as well as transfer the active anesthesia records to other work stations (e.g. induction room, operating room, recovery room etc.).
4.3. Program administration An important aspect of the smooth application in hospital is the use of independent
work stations which are not affected by network failure. For this reason, the initial separation between program application and database was retained. Configurable administrative parameters of the department, e.g. staff, drugs etc., are stored as preference data in the relational database. When changes are made, a new preferences file is created and distributed by a file distribution system (FileWave, Wave Research Inc., Berkeley CA) to the individual work stations. As soon as the program is started, the preference data are read from this file. This ensures a consistency of data at all workstations even in cases of network or server failure. Access via individual passwords ensures definitive responsibility for each anesthesia procedure. This also applies to the rejection of drugs that are subject to the German law of narcotic management (BtMG). Passwords are distributed centrally by the administrator and stored in the database after an automated 48-bit encryption process.
4.4. Integrity check The integrity check of the data set being imported into the database is carried out within the program application. In this way, the anesthesiologist records a complete data set during data entry. Time documentation was subjected to strict logical algorithms and serves as a guideline for the data recording. On one hand, the sequence and integrity of individual time points are controlled. On the other hand, numerous algorithms are run during the time documentation, e.g. checking of certain compulsory fields before transfer to the recovery room, time period recording through documentation of the start and end of periods. Thus, the documenting person receives valuable help for the important part of time documentation. The start of anesthe-
M. Benson et al. / International Journal of Medical Informatics 57 (2000) 155–164
sia presence and the end of the anesthesia procedures cannot be altered once they are entered. By default, the present system time is used when entering new events. Thus, an exact definition of anesthesia record time frames is achieved and a complete data set at the end of the anesthesia care is obtained. During the anesthesia, a list of missing data provides information on empty entries at any time. An integrity check of all compulsory fields is run before closure of the anesthesia record. This enables the anesthesiologist to correct or complete still missing data. Printout of the record is only possible after complete data entry. After the completion anesthesiological procedures and discharge of the patient, the completed record is printed out. The document is then write-protected and can no longer be altered. The network server finally imports the data set, which is stored as a text file until that point of time, into the database. Manual anesthesia records on paper can be entered into the program if necessary.
4.5. Database structure An important step in further development was the replacement of a proprietary database with a relational standard Oracle database, version 7 (Oracle Corporation). The database design, according to the principles of the relational data model, enables unambiguous statistics of the individual data fields. Evaluations and statistics can be carried out with the help of standard SQL (structured query language) tools.
161
version requires a PowerPC with 16 MB RAM and a hard disk of 500 MB. All software modules have been written using Microsoft Visual C and C++ development tools. The porting of the program from the MacOS- to the Windows-platform required a total re-programming. Computers with a high input/output efficiency, such as a Pentium-PC with PCI bus or a Macintosh workgroup server, are suitable as network servers. The patient vital data monitor is connected to the workstation through a serial interface (RS 232).
5. Routine use On January 1, 1997, NarkoData 4 was implemented hospital wide. The change from version 3.0 to version 4.0 during routine clinical work did not cause any difficulties thanks to already established software architecture. From 1997 to 1999, 66 764 anesthesia procedures at 115 workstations were recorded with the new version (1997, 21 133; 1998, 21 216; 1999, 22 284; values refer to the database inventory on February 8, 2000). Despite doubts and somewhat reserved attitudes of some staff members towards the system’s implementation, high acceptance was gained in a short period of time through intensive user service and training.
6. Discussion
4.6. Hardware requirements
6.1. Manual 6ersus automated anesthesia documentation
The program is available for IBM compatible personal computers and for Apple Macintosh PCs. A 80486DX2 processor with 16 MB RAM and a hard-disk with 250 MB is sufficient for Windows NT. The Macintosh
Increasing requirements for performance and quality assurance predetermine future problems with ordinary paper documentation. This method will no longer be sufficient to record all relevant data. A large propor-
162
M. Benson et al. / International Journal of Medical Informatics 57 (2000) 155–164
tion of manually recorded data is full of mistakes, the quality is poor [12] and direct evaluation is impossible or very time consuming and requires manual search of the records. With an automated online recording system and connection to the HIS, the presented documentation system is able to process more data fields and to keep content of the data fields consistent. The immediate data entry during the procedure and the online connection to the patient vital sign monitors lead to a comprehensive high quality documentation [12 – 14]. The system controls negligence, smaller mistakes and non-compliance of the anesthesiologist. This support, used by the documenting person, increases the validity and integrity of the records. The occurrence of artifacts still remains a partially unsolved problem in online recording. Petry [15] reported the development and application of corrective algorithms, which function as a kind of artifact filter for subsequent automatic elimination of artifacts during online data transfer. A similar solution has yet to be realized in the presented software. However, the documenting anesthesiologist can correct artifacts when recognized before the anesthesia record is closed. In the clinics’ daily routine, this method proved to be useful and sufficient. The fear that the vigilance of the anesthesiologist may be reduced by automated recording of the vital sign parameters without the necessity of a manual entry is comprehensible. However, Edsall et al. [12], as well as other authors [16–18], did not observe any influence of the anesthesia recording method on the anesthesiologist’s attention. The documentation of the anesthesia procedure is very time consuming for the responsible anesthesiologists. Some authors suggest percentages up to 20% of the total anesthesia
time [16,19,20]. The use of standard anesthesia records with pre-configured lists, HIS data import and online data recording may reduce this percentage considerably and leave the anesthesiologist more time for actual anesthesia care. Another advantage of online documentation is the improved information flow at the anesthesia work station if part of a network. The anesthesiologist can access information such as patient laboratory results directly from the HIS. Similar possibilities were described by Hohenloser et al for a patient data management system (PDMS) of an internal intensive care unit [21]. Literature provides only little information on training times. In our clinic, training is carried out by colleagues during routine work. If necessary, this is supported by the administrator who is always available. A detailed program documentation [22] is available as a web based application at any work station.
6.2. The database The first evaluations after successful implementation of version 3 showed the limited spectrum of the non-relational database query options. This did not at all meet the high expectations and did not justify the investment. Similar problems were reported by Zbinden et al. [23]. For them, this was one of the reasons to change the system. The switch to a relational database with a centralized architecture of the database system was the most important step in the program development and finally lead to a significant increase in performance. The crucial point of the system is the generation of data via pre-configured choice lists rather than free text entry. The free text entry of drug names for example inevitably results in a number of names for the same product
M. Benson et al. / International Journal of Medical Informatics 57 (2000) 155–164
which render the evaluation process much more difficult [24]. Another advantage of the system is the optional use of standard SQL tools for evaluation. A similar standard that offers the possibility for an extensive documentation and evaluation of complete and close data sets cannot be achieved by manual paper based systems [12– 14,25]. With a comparatively small effort, statistics and evaluations for administration, training, research and departmental issues can be far easier designed and are more reliable. The extent of the system’s suitability for scientific research will be investigated in the near future by increasing system use for scientific queries. Decisive advantages are the large number of cases and the relatively high data quality [5]. In contrast to automatically readable anaesthetic records, where limitation of recorded content in favor of directly evaluable data makes sense [7], the fear of creating ‘data cemeteries’ with more comprehensive automated recording systems can be understood. However, this attitude is not futureoriented. Only a broad spectrum of data recording allows various requirements to be fulfilled without redundancy. It sets the basis for a larger anesthesiological database, which may lead to an efficient expert system in the future.
6.3. Future de6elopments The specialized program NarkoData records only the perioperative part in anesthesia. Numerous other tasks needing to be recorded, such as central venous catheterization unrelated to surgical procedures, analgesic catheter ward round or treatment in the out-patient analgesic clinic, cannot be documented with the system in its current form. Furthermore, the integration of the postoper-
163
ative visit into the documenting system should be made possible at each ward. This requires a reliable connection to the HIS and other subsystems which can only be achieved via standard interfaces, such as HL7, or via a direct link to the database. At the moment, trials are being run in two departments with a supplementary software product that will be used primarily for communication (HL-7) and quick database access. This system is designed for the documentation of anesthesia care not directly related to an anesthesia.
7. Conclusion The joint cooperation of an anesthesiological department at a university hospital and a software company proved to be a suitable way to design an online documentation system. The realization of the main points of the requirement document enabled us to develop an anesthesia documentation system that supplies an exact, comprehensive and legible anesthesia record, along with an ergonomic data presentation for statistical use and an economical archive access.
References [1] N.P. Hirsch, G.B. Smith, Harvey Cushing: his contribution to anaesthesia, Anaesth. Analg. 65 (1986) 288 – 293. [2] B. Schwilk, J. Ha¨hnel, L. Friess, S. Blessing, W. Friesdorf, Ergebnisse zur Verfahrensentwicklung und zum personellen Aufwand im Zusammenhang mit der Einfu¨hrung qualita¨tssichernder Maßnahmen, Ana¨sth. Intensivmed. 35 (1994) 349– 356. [3] F.W. Ahnefeld, Empfehlungen der DGAI zur Qualita¨tssicherung: Kerndatensatz Ana¨sthesie, Ana¨sth. Intensivmed. 34 (1993) 330. [4] A. Bach, M. Bauer, konomische Aspekte in der Ana¨sthesie, Ana¨sthesiol. Intensivmed. Notfallmed. Schmerzther. 33 (1998) 135 – 149.
164
M. Benson et al. / International Journal of Medical Informatics 57 (2000) 155–164
[5] DGAI-Kommission, Qualita¨tssicherung und Datenverarbeitung in der Ana¨sthesie: Kerndatensatz Qualita¨tssicherung in der Ana¨sthesie, Ana¨sth. Intensivmed. 34 (1993) 331–335. [6] H.W. Opderbecke, W. Weißauer, Eine Empfehlung zur einheitlichen Protokollierung von Ana¨sthesieverfahren, Ana¨sth. Intensivmed. 30 (1989) 189. [7] A. Tecklenburg, Form und Inhalt eines EDVgerechten Ana¨sthesieprotokolls, Ana¨sth. Intensivmed. 34 (1993) 93–100. [8] P.M. Osswald, DGAI-Kommission EDV in der Ana¨sthesie: Empfehlungen zu Grundvorstellungen u¨ber die Inhalte des Ana¨sthesieprotokolls und des pra¨operativen Zustandsprotokolls Teil I: Katalog der Inhalte des pra¨operativen Zustandsprotokolls, Ana¨sth. Intensivmed. 30 (1989) 205–207. [9] P.M. Osswald, DGAI-Kommission EDV in der Ana¨sthesie: Empfehlungen zu Grundvorstellungen u¨ber die Inhalte des Ana¨sthesieprotokolls und des pra¨operativen Zustandsprotokolls Teil II: Katalog der Inhalte des Ana¨sthesieprotokolls, Ana¨sth. Intensivmed. 30 (1989) 239–241. [10] W. Heinrichs, P.M. Osswald, A. Tecklenburg, Information zu Computerprogrammen im Bereich Ana¨sthesie und Intensivmedizin, Ana¨sth. Intensivmed. 33 (1992) 73–76. [11] Thieme Software, Thieme NarkoData, Handbuch zur Version 3.0. [12] D.W. Edsall, P. Deshane, C. Giles, B. Sloan, Computerized patient anesthesia records: less time and better quality than manually produced anesthesia records, J. Clin. Anaesth. 5 (1993) 275–283. [13] J.P. Abenstein, C.B. De Vos, M.D. Abel, S. Tarhan, Eight year’s experience with automated anesthesia record keeping: lessons learned — new directions taken, Int. J. Clin. Monit. Comput. 9 (1992) 117–129. [14] D.N. Thrush, Are automated anaesthesia records better?, J. Clin. Anesth. 4 (1992) 386–393.
.
[15] A. Petry, On-line Aufzeichnung von Monitordaten, Anaesthesist 44 (1995) 818 – 825. [16] J. Allard, R. Dzwonezyk, D. Yablok, F.E. Block, J.S. McDonald, Effect of automatic record keeping on vigilance and record keeping time, Br. J. Anesth. 74 (1995) 619 – 626. [17] V. Lanza, Automatic record keeping in anaesthesia — a 9-year Italian experience, Int. J. Clin. Monit. Comput. 13 (1996) 35 – 43. [18] R.G. Loeb, Manual record keeping is not necessary for anesthesia vigilance, J. Clin. Monit. 11 (1995) 9 – 13. [19] R.M. Gardner, O. Prakash, Challenges and opportunities for computerizing the anesthesia record, J. Clin. Anesth. 6 (1994) 333 – 341. [20] J.G.C. Lerou, R. Dirksen, M. van Daele, G.M.M. Nghurs, Automated charting of physiological data variables in anesthesia: a quantitative comparison of automated versus handwritten anesthesia records, J. Clin. Monit. 4 (1988) 37 – 47. [21] J.H. Hohenloser, F. Pu¨rner, PADS (patient archiving and documentation system): a computerized patient record with educational aspects, Int. J. Clin. Monit. Comput. 9 (1992) 71 – 84. [22] NarkoData-Dokumentation, Version 4, http:// www.prologic.de/. [23] A.M. Zbinden, H. Rothenbu¨hler, B. Ha¨berli, Datenerfassung in der Ana¨sthesie, Ana¨sthesist 46 (1997) 496 – 503. [24] D.W. Edsall, B.R. Jones, N.T. Smith, The anesthesia database, the automated record, and the quality assurance process, Int. Anaesthesiol. Clin. 2 (1992) 71 – 92. [25] A. Kari, E. Ruokonen, J. Takala, Comparison of acceptance and performance of automated and manual data management systems in intensive care, Int. J. Clin. Monit. Comput. 7 (1990) 157– 162.
www.elsevier.com/locate/ijmedinf
Clinical and practical requirements of online software for anesthesia documentation—an experience report Matthias Benson a,*, Axel Junger b, Lorenzo Quinzio a, Carsten Fuchs a, Gregor Sciuk a, Achim Michel b, Kurt Marquardt b, Gunter Hempelmann a a
Department of Anesthesiology and Intensi6e Care Medicine, Justus-Liebig Uni6ersity, Rudolf-Buchheim Strasse, 7, 35392 Giessen, Germany b Department of Medical and Administrati6e Data Processing, Justus-Liebig Uni6ersity, Klinikstraße 23, 35392 Giessen, Germany Received 18 May 1999; received in revised form 2 March 2000; accepted 28 March 2000
Abstract The aim of this paper is the presentation of a new version of the anesthesia documentation software, NarkoData, that has been used in routine clinical work in our department as part of an anesthesia information management system (AIMS) since 1995. The performance of this software is presented along with requirements for future development of such a system. The originally used version, NarkoData 3.0, is an online anesthesia documentation software established by the software company ProLogic GmbH. It was primarily developed as a disk-based system for the MacOS operating system (Apple Computer Inc.). Based on our routine experience with the system, a catalogue of requirements was developed that concentrated on improvement in the sequence of work, administration and data management. In 1996, the concepts developed in our department, in close co-operation with medical personnel and the software company, led to a considerable enlargement of the program functions and the subsequent release of a new version of NarkoData. Since 1997, more than 20 000 anesthesia procedures have been recorded annually with this new version at 115 decentralized work stations at our university hospital. © 2000 Elsevier Science Ireland Ltd. All rights reserved. Keywords: Anesthesiology; Automatic anesthesia record; Personal computer; Documentation; Quality assurance
1. Introduction As early as 1895, a formal documentation of an anesthesia procedure was made [1]. * Corresponding author. Tel.: + 49-641-9944494; fax: +49641-9944499. E-mail address: [email protected] (M. Benson).
Since then, anesthesia documentation has been developed continuously and is now an integral part of anesthesiology. At present, the documentation is mainly done on paper. Computer entry of this data is supported by machine-readable forms and/ or done manually after the anesthesia. These methods are very time-consuming [2] and do
1386-5056/00/$ - see front matter © 2000 Elsevier Science Ireland Ltd. All rights reserved. PII: S 1 3 8 6 - 5 0 5 6 ( 0 0 ) 0 0 0 5 9 - 9
156
M. Benson et al. / International Journal of Medical Informatics 57 (2000) 155–164
not meet increasing quality requirements [3– 7]. Another, presently less-used method of documentation, is automated online recording during anesthesia, including automatic data transfer from patient monitoring systems and connection to the hospital information system (HIS). For effective use, an anesthesia documentation software should provide user-friendly, ergonomic interfaces for data entry and extensive evaluation of data. The criteria of the German society for Anesthesiology and Intensive Care Medicine (DGAI) [5,8,9] must be met. Also, safety against failure and the administration of the system during the hospital routine working hours must be ensured. In 1994, the Department of Anesthesiology and Intensive Care Medicine at the JustusLiebig University, Giessen, decided to implement an anesthesia information management system (AIMS). The requirements for such documentation software were put down in a developer’s document. At that time, no adequate commercial product was available to fulfil the high performance requirements, so we signed a co-operation contract with a software company for the development of the program for clinical routine use. Now, after 4 years of experience with the system, the time has come for a critical review and evaluation of the extent of our attaining the objectives.
2. The last version The version NarkoData 3.0 of the anesthesia documentation software, installed in 1995, had initially been developed by the Department of Anesthesiology, Intensive Care Medicine and Analgesia at the University of Bochum (Klinik Bergmannsheil, Head of Department: Professor Dr M. Zenz) in co-operation with the company ProLogic GmbH
(Erkrath, Germany) [10]. The original program recorded medical data during anesthesia and ran as a local application at the workstation. The software was disk-based and designed for the operating system MacOS (Apple Computer, Inc., Cupertino CA) [11]. Completed protocol files were imported into a proprietary database (OMNIS Software, San Bruno CA). Due to the simple database structure, it offered only limited query options. Recording of patient data was carried out with: free text entries multiple-choice fields choice lists and a standard query for the core data pool defined by the DGAI [5]. Many parameters (ventilation, diagnosis) could only be documented once, although they could change over time during the period of anesthesia. Duration of anesthesia procedures was recorded only with the help of very few plausibility checks. Specific data (department information, medication, staff etc.) were read at the start of a new record from a preferences file, which was created by a special administration tool (SuperNarko). The program was not designed to run as part of a network, e.g. it was impossible to import data from the hospital information system (HIS). Administrative effort for the determination of standards, data evaluation and user service in a decentralized hospital was considerably high.
3. Requirements After a thorough check of the options provided by the initial version of the system, a set of requirements was set up for further development. Of particular value was the ex-
M. Benson et al. / International Journal of Medical Informatics 57 (2000) 155–164
Table 1 Main requirements for an online anesthesia documentation software, developed by the Department for Anesthesiology and Intensive Care Medicine of the Justus-Liebig University, Giessen (1) Adaptation to the anesthesiological process Large medical data fields and structure of data input after the work process Permanently visible graphic user interface with time axis (worksheet) Ergonomic data input via mouse, keyboard and shortcuts (hot keys) Representation of the time-dependent parameters on the worksheet with the possibility of multiple documentation Representation of the time-independent parameters in dialogue windows with overview display Data fields for a preoperative and a postoperative visit Sufficient space for comments on the individual parameters Use of the online data transfer from medical equipment (patient monitoring systems, analysis tools) Data import from all available computer systems (HIS, subsystems of other departments, anesthesiological data base) via a standard interface (e.g. HL-7) Help functions with a completeness check available anytime No additional time needed Fast program performance and an adequate process adapted screen presentation (2) Program administration Fixed data structure with configurable medical contents Consequent sequence and logical cheeks of the anesthesia times as guideline for complete documentation Completeness controls for the obligatory entry fields and logical algorithms to check the entered data Password protection of the anesthesiologist in charge High data safety through periodical safety copies High reliability through local program application Combination of network and stand-alone architecture Small administrative effort for installation Small administrative effort for the maintenance of the programs base data Compatible transfer to Windows NT and vice versa
157
(3) Anesthesiological database Use of a relational standard database Clear data structure, division into base tables and process tables Complete and consistent data sets Access to and print of previous anesthesia records at the workstation Export of data from previous anesthesia records into current records Availability of all data for evaluation with a standard SQL tool
perience of daily administration and data management gained from the existing patient data management system (PDMS) Emtek 2000 (Siemens AG, Munich, Germany) at our operative intensive care unit (ICU). In a brief abstract of the 150 page requirement document, the main demands are highlighted: The three most important aspects were an improved adaptation to the course of the anesthetic procedure, simplification of the program administration and improvement of the database structure. Individual points of these three aspects are shown in Table 1. Since then, this new version, NarkoData 4, has been further developed following the requirements and the experience from using NarkoData version 3.0. The new program version was improved continuously. Suggestions from the users were rated and realized whenever possible.
4. The actual version
4.1. Adaptation to anesthesia working procedures The main function expansions of Version 4 are listed in Table 2. For more flexibility, this version is available for MacOS and MS-Windows based PCs. The worksheet with its graphic user interface (Fig. 1) corresponds to the conventional
Fig. 1. The worksheet is the central graphic user interface for displaying and recording time-dependant data. The time-dependent data windows provide a quick overview.
158 M. Benson et al. / International Journal of Medical Informatics 57 (2000) 155–164
M. Benson et al. / International Journal of Medical Informatics 57 (2000) 155–164
Table 2 Decisive functional expansions of the new program version 4 (1) Adaptation to the anesthesiological process Inclusion of previously missing data fields for medical issues Adjusting of time recording, adaptation to the work process Improvement of the graphic user interface ergonomics Import option for data from previous anesthesia records into the current record Expansion of the software drivers for automatic data transfer from additional patient monitoring systems Import option for HIS data via ASCII text, Apple-Events and DDE (Windows) Redesign and enhancement of the printout Improvement of the help functions and overview windows (2) Program administration Improvement of the administration and consistency of the program’s base data Expansion of the network functions of the file management New programming of a compatible Windows NT 6ersion (3) Anesthesiological database Reorganization and data transfer of the data from a hierarchical to a relational standard database Improvement of data consistency through implementation of logical checking algorithms during data input Improvement of data quality through import mechanisms from other subsystems Clear display and printout of previous anesthesia records
paper report. It offers a permanently visible documentation of the complete procedure including the recovery room. Differences in color guarantee clear recognition. The ergonomic and intuitive entry of data during the procedure, e.g. vital data, drugs given and anesthesia times, is directly performed by mouse click. Additional time-dependent parameters can be entered or
159
presented at the desired time in dialogue windows opened by simple double-click. The sequence of data entry is adapted to the actual course of the anesthetic procedure. Pre-configured sequences of dialogue windows, plenty of pre-configured choice lists and logical checks via cross references support a fast and complete documentation. Hence, the required time for staff training is considerably reduced (approximately one working day). The system records time-independent data fields (administrative data, preoperative diagnoses, ICD and ICPM keys, measures and biometrical data) and time-dependent parameters such as vital and respiratory values, drugs given, laboratory data, adverse events, etc. The data fields of the core data set of the DGAI [5] were originally developed for paper documentation and could now be adjusted to accommodate the possibilities of digital online recording. The coding of diagnoses and surgery data follows common coding keys (e.g. ICD, ICPM) stored in the database. An overview window allows an instant orientation in the contents of the dialogue windows at any time (Fig. 1). This is especially helpful for patient transfer. Call of pop-up windows or data entry fields via hot keys or mouse is executed immediately; server waiting times do not affect the use of the system. The program offers the possibility to record preoperative data with a laptop computer at the patient’s bedside in the ward. This pre-operative visit can be printed after an automatic integrity check is carried out, thus providing the ward’s staff with written instructions. This information is made available to the anesthesiologist on the day of the procedure. A similar documentation for the postoperative visit has not yet been realized. If pre-configured, standardized anesthesia record forms are used, the administrative,
160
M. Benson et al. / International Journal of Medical Informatics 57 (2000) 155–164
preoperative and surgery-related data can be entered in 5–10 min. The importation of patient core data and laboratory data from previous anesthesia records cuts the time to 3 – 5 min. The content of the pre-configured anesthesia record forms is usually determined by the doctor in charge.
4.2. System architecture The program itself runs as a local application. Every 2 minutes, two independent safety copies of the anesthesia protocol are recorded in plain text format, one on the local harddisk and the other on another data storage medium. Because the safety copies are made consecutively on two different data storage media, the maximum loss of data is cut to one storage interval (2 min). If the program is running on a network, the second data storage medium is set by default to a server providing all connected clients with record files. Computers, which are not connected to the network, use a floppy disk as second data storage medium. These disks also ensure data exchange between the individual work stations, in case of network or server failure, by loading the anesthesia protocol stored on the floppy disk and automatically saving it on the server when the computer is reconnected to the network. Thus, a mixed architecture of networked and independent computers is possible. Within the computer network, it is possible to import data from the HIS (via AppleEvents or DDE), as well as transfer the active anesthesia records to other work stations (e.g. induction room, operating room, recovery room etc.).
4.3. Program administration An important aspect of the smooth application in hospital is the use of independent
work stations which are not affected by network failure. For this reason, the initial separation between program application and database was retained. Configurable administrative parameters of the department, e.g. staff, drugs etc., are stored as preference data in the relational database. When changes are made, a new preferences file is created and distributed by a file distribution system (FileWave, Wave Research Inc., Berkeley CA) to the individual work stations. As soon as the program is started, the preference data are read from this file. This ensures a consistency of data at all workstations even in cases of network or server failure. Access via individual passwords ensures definitive responsibility for each anesthesia procedure. This also applies to the rejection of drugs that are subject to the German law of narcotic management (BtMG). Passwords are distributed centrally by the administrator and stored in the database after an automated 48-bit encryption process.
4.4. Integrity check The integrity check of the data set being imported into the database is carried out within the program application. In this way, the anesthesiologist records a complete data set during data entry. Time documentation was subjected to strict logical algorithms and serves as a guideline for the data recording. On one hand, the sequence and integrity of individual time points are controlled. On the other hand, numerous algorithms are run during the time documentation, e.g. checking of certain compulsory fields before transfer to the recovery room, time period recording through documentation of the start and end of periods. Thus, the documenting person receives valuable help for the important part of time documentation. The start of anesthe-
M. Benson et al. / International Journal of Medical Informatics 57 (2000) 155–164
sia presence and the end of the anesthesia procedures cannot be altered once they are entered. By default, the present system time is used when entering new events. Thus, an exact definition of anesthesia record time frames is achieved and a complete data set at the end of the anesthesia care is obtained. During the anesthesia, a list of missing data provides information on empty entries at any time. An integrity check of all compulsory fields is run before closure of the anesthesia record. This enables the anesthesiologist to correct or complete still missing data. Printout of the record is only possible after complete data entry. After the completion anesthesiological procedures and discharge of the patient, the completed record is printed out. The document is then write-protected and can no longer be altered. The network server finally imports the data set, which is stored as a text file until that point of time, into the database. Manual anesthesia records on paper can be entered into the program if necessary.
4.5. Database structure An important step in further development was the replacement of a proprietary database with a relational standard Oracle database, version 7 (Oracle Corporation). The database design, according to the principles of the relational data model, enables unambiguous statistics of the individual data fields. Evaluations and statistics can be carried out with the help of standard SQL (structured query language) tools.
161
version requires a PowerPC with 16 MB RAM and a hard disk of 500 MB. All software modules have been written using Microsoft Visual C and C++ development tools. The porting of the program from the MacOS- to the Windows-platform required a total re-programming. Computers with a high input/output efficiency, such as a Pentium-PC with PCI bus or a Macintosh workgroup server, are suitable as network servers. The patient vital data monitor is connected to the workstation through a serial interface (RS 232).
5. Routine use On January 1, 1997, NarkoData 4 was implemented hospital wide. The change from version 3.0 to version 4.0 during routine clinical work did not cause any difficulties thanks to already established software architecture. From 1997 to 1999, 66 764 anesthesia procedures at 115 workstations were recorded with the new version (1997, 21 133; 1998, 21 216; 1999, 22 284; values refer to the database inventory on February 8, 2000). Despite doubts and somewhat reserved attitudes of some staff members towards the system’s implementation, high acceptance was gained in a short period of time through intensive user service and training.
6. Discussion
4.6. Hardware requirements
6.1. Manual 6ersus automated anesthesia documentation
The program is available for IBM compatible personal computers and for Apple Macintosh PCs. A 80486DX2 processor with 16 MB RAM and a hard-disk with 250 MB is sufficient for Windows NT. The Macintosh
Increasing requirements for performance and quality assurance predetermine future problems with ordinary paper documentation. This method will no longer be sufficient to record all relevant data. A large propor-
162
M. Benson et al. / International Journal of Medical Informatics 57 (2000) 155–164
tion of manually recorded data is full of mistakes, the quality is poor [12] and direct evaluation is impossible or very time consuming and requires manual search of the records. With an automated online recording system and connection to the HIS, the presented documentation system is able to process more data fields and to keep content of the data fields consistent. The immediate data entry during the procedure and the online connection to the patient vital sign monitors lead to a comprehensive high quality documentation [12 – 14]. The system controls negligence, smaller mistakes and non-compliance of the anesthesiologist. This support, used by the documenting person, increases the validity and integrity of the records. The occurrence of artifacts still remains a partially unsolved problem in online recording. Petry [15] reported the development and application of corrective algorithms, which function as a kind of artifact filter for subsequent automatic elimination of artifacts during online data transfer. A similar solution has yet to be realized in the presented software. However, the documenting anesthesiologist can correct artifacts when recognized before the anesthesia record is closed. In the clinics’ daily routine, this method proved to be useful and sufficient. The fear that the vigilance of the anesthesiologist may be reduced by automated recording of the vital sign parameters without the necessity of a manual entry is comprehensible. However, Edsall et al. [12], as well as other authors [16–18], did not observe any influence of the anesthesia recording method on the anesthesiologist’s attention. The documentation of the anesthesia procedure is very time consuming for the responsible anesthesiologists. Some authors suggest percentages up to 20% of the total anesthesia
time [16,19,20]. The use of standard anesthesia records with pre-configured lists, HIS data import and online data recording may reduce this percentage considerably and leave the anesthesiologist more time for actual anesthesia care. Another advantage of online documentation is the improved information flow at the anesthesia work station if part of a network. The anesthesiologist can access information such as patient laboratory results directly from the HIS. Similar possibilities were described by Hohenloser et al for a patient data management system (PDMS) of an internal intensive care unit [21]. Literature provides only little information on training times. In our clinic, training is carried out by colleagues during routine work. If necessary, this is supported by the administrator who is always available. A detailed program documentation [22] is available as a web based application at any work station.
6.2. The database The first evaluations after successful implementation of version 3 showed the limited spectrum of the non-relational database query options. This did not at all meet the high expectations and did not justify the investment. Similar problems were reported by Zbinden et al. [23]. For them, this was one of the reasons to change the system. The switch to a relational database with a centralized architecture of the database system was the most important step in the program development and finally lead to a significant increase in performance. The crucial point of the system is the generation of data via pre-configured choice lists rather than free text entry. The free text entry of drug names for example inevitably results in a number of names for the same product
M. Benson et al. / International Journal of Medical Informatics 57 (2000) 155–164
which render the evaluation process much more difficult [24]. Another advantage of the system is the optional use of standard SQL tools for evaluation. A similar standard that offers the possibility for an extensive documentation and evaluation of complete and close data sets cannot be achieved by manual paper based systems [12– 14,25]. With a comparatively small effort, statistics and evaluations for administration, training, research and departmental issues can be far easier designed and are more reliable. The extent of the system’s suitability for scientific research will be investigated in the near future by increasing system use for scientific queries. Decisive advantages are the large number of cases and the relatively high data quality [5]. In contrast to automatically readable anaesthetic records, where limitation of recorded content in favor of directly evaluable data makes sense [7], the fear of creating ‘data cemeteries’ with more comprehensive automated recording systems can be understood. However, this attitude is not futureoriented. Only a broad spectrum of data recording allows various requirements to be fulfilled without redundancy. It sets the basis for a larger anesthesiological database, which may lead to an efficient expert system in the future.
6.3. Future de6elopments The specialized program NarkoData records only the perioperative part in anesthesia. Numerous other tasks needing to be recorded, such as central venous catheterization unrelated to surgical procedures, analgesic catheter ward round or treatment in the out-patient analgesic clinic, cannot be documented with the system in its current form. Furthermore, the integration of the postoper-
163
ative visit into the documenting system should be made possible at each ward. This requires a reliable connection to the HIS and other subsystems which can only be achieved via standard interfaces, such as HL7, or via a direct link to the database. At the moment, trials are being run in two departments with a supplementary software product that will be used primarily for communication (HL-7) and quick database access. This system is designed for the documentation of anesthesia care not directly related to an anesthesia.
7. Conclusion The joint cooperation of an anesthesiological department at a university hospital and a software company proved to be a suitable way to design an online documentation system. The realization of the main points of the requirement document enabled us to develop an anesthesia documentation system that supplies an exact, comprehensive and legible anesthesia record, along with an ergonomic data presentation for statistical use and an economical archive access.
References [1] N.P. Hirsch, G.B. Smith, Harvey Cushing: his contribution to anaesthesia, Anaesth. Analg. 65 (1986) 288 – 293. [2] B. Schwilk, J. Ha¨hnel, L. Friess, S. Blessing, W. Friesdorf, Ergebnisse zur Verfahrensentwicklung und zum personellen Aufwand im Zusammenhang mit der Einfu¨hrung qualita¨tssichernder Maßnahmen, Ana¨sth. Intensivmed. 35 (1994) 349– 356. [3] F.W. Ahnefeld, Empfehlungen der DGAI zur Qualita¨tssicherung: Kerndatensatz Ana¨sthesie, Ana¨sth. Intensivmed. 34 (1993) 330. [4] A. Bach, M. Bauer, konomische Aspekte in der Ana¨sthesie, Ana¨sthesiol. Intensivmed. Notfallmed. Schmerzther. 33 (1998) 135 – 149.
164
M. Benson et al. / International Journal of Medical Informatics 57 (2000) 155–164
[5] DGAI-Kommission, Qualita¨tssicherung und Datenverarbeitung in der Ana¨sthesie: Kerndatensatz Qualita¨tssicherung in der Ana¨sthesie, Ana¨sth. Intensivmed. 34 (1993) 331–335. [6] H.W. Opderbecke, W. Weißauer, Eine Empfehlung zur einheitlichen Protokollierung von Ana¨sthesieverfahren, Ana¨sth. Intensivmed. 30 (1989) 189. [7] A. Tecklenburg, Form und Inhalt eines EDVgerechten Ana¨sthesieprotokolls, Ana¨sth. Intensivmed. 34 (1993) 93–100. [8] P.M. Osswald, DGAI-Kommission EDV in der Ana¨sthesie: Empfehlungen zu Grundvorstellungen u¨ber die Inhalte des Ana¨sthesieprotokolls und des pra¨operativen Zustandsprotokolls Teil I: Katalog der Inhalte des pra¨operativen Zustandsprotokolls, Ana¨sth. Intensivmed. 30 (1989) 205–207. [9] P.M. Osswald, DGAI-Kommission EDV in der Ana¨sthesie: Empfehlungen zu Grundvorstellungen u¨ber die Inhalte des Ana¨sthesieprotokolls und des pra¨operativen Zustandsprotokolls Teil II: Katalog der Inhalte des Ana¨sthesieprotokolls, Ana¨sth. Intensivmed. 30 (1989) 239–241. [10] W. Heinrichs, P.M. Osswald, A. Tecklenburg, Information zu Computerprogrammen im Bereich Ana¨sthesie und Intensivmedizin, Ana¨sth. Intensivmed. 33 (1992) 73–76. [11] Thieme Software, Thieme NarkoData, Handbuch zur Version 3.0. [12] D.W. Edsall, P. Deshane, C. Giles, B. Sloan, Computerized patient anesthesia records: less time and better quality than manually produced anesthesia records, J. Clin. Anaesth. 5 (1993) 275–283. [13] J.P. Abenstein, C.B. De Vos, M.D. Abel, S. Tarhan, Eight year’s experience with automated anesthesia record keeping: lessons learned — new directions taken, Int. J. Clin. Monit. Comput. 9 (1992) 117–129. [14] D.N. Thrush, Are automated anaesthesia records better?, J. Clin. Anesth. 4 (1992) 386–393.
.
[15] A. Petry, On-line Aufzeichnung von Monitordaten, Anaesthesist 44 (1995) 818 – 825. [16] J. Allard, R. Dzwonezyk, D. Yablok, F.E. Block, J.S. McDonald, Effect of automatic record keeping on vigilance and record keeping time, Br. J. Anesth. 74 (1995) 619 – 626. [17] V. Lanza, Automatic record keeping in anaesthesia — a 9-year Italian experience, Int. J. Clin. Monit. Comput. 13 (1996) 35 – 43. [18] R.G. Loeb, Manual record keeping is not necessary for anesthesia vigilance, J. Clin. Monit. 11 (1995) 9 – 13. [19] R.M. Gardner, O. Prakash, Challenges and opportunities for computerizing the anesthesia record, J. Clin. Anesth. 6 (1994) 333 – 341. [20] J.G.C. Lerou, R. Dirksen, M. van Daele, G.M.M. Nghurs, Automated charting of physiological data variables in anesthesia: a quantitative comparison of automated versus handwritten anesthesia records, J. Clin. Monit. 4 (1988) 37 – 47. [21] J.H. Hohenloser, F. Pu¨rner, PADS (patient archiving and documentation system): a computerized patient record with educational aspects, Int. J. Clin. Monit. Comput. 9 (1992) 71 – 84. [22] NarkoData-Dokumentation, Version 4, http:// www.prologic.de/. [23] A.M. Zbinden, H. Rothenbu¨hler, B. Ha¨berli, Datenerfassung in der Ana¨sthesie, Ana¨sthesist 46 (1997) 496 – 503. [24] D.W. Edsall, B.R. Jones, N.T. Smith, The anesthesia database, the automated record, and the quality assurance process, Int. Anaesthesiol. Clin. 2 (1992) 71 – 92. [25] A. Kari, E. Ruokonen, J. Takala, Comparison of acceptance and performance of automated and manual data management systems in intensive care, Int. J. Clin. Monit. Comput. 7 (1990) 157– 162.