original article Wien Klin Wochenschr [Suppl] DOI 10.1007/s00508-015-0794-7
Lightweight application for generating clinical research information systems: MAGIC Brane Leskošek · Marjan Pajntar
Received: 22 January 2013 / Accepted: 20 April 2015 © Springer-Verlag Wien 2015
Summary Background Our purpose was to build and test a lightweight solution for generating clinical research information systems (CRIS) that would allow non-IT professionals with basic knowledge of computer usage to quickly define and build a ready-to-use, safe and secure webbased clinical research system for data management. We use the acronym MAGIC (Medical Application Generator InteraCtive) for the system. Methods The generated CRIS should be very easy to build and use, so a common LAMP (Linux Apache MySQL Perl) platform was used, which also enables short development cycles. The application was built and tested using eXtreme Programming (XP) principles by a small development team consisting of one informatics specialist, one physician and one graphical designer/ programmer. Results The parameter and graphical user interface (GUI) definitions for the CRIS can be made by non-IT professionals using an intuitive English-language-like formalism called application definition language (ADL). From these definitions, the MAGIC builds an end-user CRIS that can be used on a wide variety of platforms (from B. Leskošek, PhD, EE () Faculty of Medicine, Institute for Biostatistics and Medical Informatics, University of Ljubljana, Vrazov trg 2, 1000 Ljubljana, Slovenia e-mail: [email protected] B. Leskošek, PhD, EE Faculty of Medicine, University of Maribor, Taborska ulica 8, 2000 Maribor, Slovenia M. Pajntar, PhD, MD Faculty of Medicine, University of Ljubljana, Vrazov trg 2, 1000 Ljubljana, Slovenia
13
standard workstations to hand-held devices). A working example of a national health-care-quality assessment program is presented to illustrate this process. Conclusion The lightweight application for generating CRIS (MAGIC) has proven to be useful for both clinical and analytical users in real working environment. To achieve better performance and interoperability, we are planning to recompile the application using XML schemas (XSD) in HL7 CDA or openEHR archetypes formats used for parameters definition and for data interchange between different information systems. Keywords Information system · Clinical trial · Electronic health record · Clinical research informatics · Interoperability
Introduction Information technology (IT) has become a key resource for research institutions, providing services such as hardware, software and network maintenance, as well as data management services [1, 2]. With the emerging new technologies, the execution of wide range of new different clinical trials becomes possible. New laboratory equipment and new analytical methods makes possible to do many different new examinations and gives us a lot of new data for each patient or sample. A lot of these different clinical trials/studies [3, 4], which are done every day, need (usually electronic) data management (planning, acquisition, validation, quality and analysis). Even though a rapid development of new IT makes data collection more easy, safe, secure and reliable, the connections between (and among) all these data generated by new equipment and methods and between patients/ samples becomes more and more complex. That arises the need for more complex data management systems that will assure structured quality data collection that is
Lightweight application for generating clinical research information systems: MAGIC
1
original article
needed for quality analyses and results. Because of the complexity of data connections and definitions, the data management still takes a lot of time and effort even in cases when all trial parameters are already defined in advance. Sometimes that takes even majority of trial’s time. The main challenge is that the available solutions (either commercial or open source) for data management [5–16] need a professional information specialist/programmer and have very steep and long learning curve, mainly because the IT solutions tries to cover as wide functionalities and options as possible. The needed “usual” use of IT professionals can be also concluded from [2]. However, there is a lack of simple and not so sophisticated clinical research information systems (CRIS) data management tools that would enable the basic functionality of safe, secure and reliable quality data management and close the gap between simple and complex data management systems. That is especially evident in smaller clinical trials where usually the professional information specialist is not available or it would be too expensive to have it in a research team. But professional information specialist is actually needed for using more complex solutions [12–16]. This situation many times results in poor research (data and quality management) information systems that leads to bad data quality, (long) time delays, unreliable analyses, large expenses, hard-to-adapt IT systems, etc. The purpose of this investigation was to build and test a specific lightweight solution for generation of CRIS (MAGIC—Medical Application Generator InteraCtive). The aims of the MAGIC are as follows: ●● to enable rapid parameters/data definitions by nonIT professionals, which have some very basic knowledge about computer usage and are usually principal medical investigators such as medical specialists— physicians, pharmacologists, pharmacists, molecular biologists …; ●● that only very widespread and basic computer tools are needed for parameters/data and user interface definitions, e.g. MS Excel, any (Unicode UTF-8 enabled) ASCII editor (e.g. Notepad, Notepad ++ …); ●● to get a ready-to-use safe and secure CRIS for fast data management and with low (establishment and maintenance) costs; ●● to have a very intuitive generated CRIS application usage where none or only very short training time is needed for end-users; ●● that no installation or changes or adaptations of endusers’ workstations/computers is needed. The generated CRIS with basic functionalities should be used from as many different computers/operation systems as possible and should run within existing organizational information systems without the need for any changes, adoption or installation of workstations software;
●● that the usage from different dislocated units (e.g. laboratories, institutions …) should be integrated as the basic functionality in the generated CRIS; ●● that the generated CRIS can be easily upgraded with more sophisticated functionalities and different graphical design even though for that a trained and skilled information specialist/programmer is needed.
Methods The CRIS generated using MAGIC should be very easy to build and use and maintained using short development cycles. The primary principle of building method for CRIS using MAGIC was eXtreme Programming (XP) [17]. We use that principle mainly for the end-users application’s (“generated CRIS”) parameters definitions, installation and (intuitive) usage. CRIS application is web based because of the world’s internet penetration with well-tested security mechanisms (like SSL cyphering algorithms …). All communication and data transfer is done only via http(s) protocols. To follow previously mentioned lightweight requests and always present need for fast development cycles, a very common open source LAMP (Linux Apache MySQL Perl) platform was used. The generation tool was built and tested using XP principles [17] in very small development team consisting from one informatics specialist, one physician and one graphical designer/programmer. To allow usage on a variety of different platforms and from many different places regardless of the internet connection bandwidth, the end-user’s CRIS application should exploit only basic web technologies.
Platforms used for generation tool and CRIS CRIS application exploits only basic web technologies like (X)HTML, JavaScript, HTML CGI, POST and GET methods and cascading style sheets (CSS) for (X)HTML documents’ visual presentation/graphical design. Clinical trials functionality is at least in a (first) phase when data are collected, much more important than application’s graphical design, especially because graphical design could limit the spread of (CRIS) application usage. “Fancy” technologies like Adobe Flash, Silverlight and similar were avoided because they usually adds only more “appealing” (also depending on user’s taste) visual appearance but no or just very limited added functionality. And last but not least important goal was also to develop a robust and operating system-independent solution. The supported and tested client-side platforms are UNIX-like (Linux, BSD, Mac OS X …), MS Windows, iOS, Android, VMS … with many different XHTML compatible browsers like (newer versions of ) Mozilla Firefox, MS Internet Explorer, Google Chrome, Apple Safari, Opera ….
2 Lightweight application for generating clinical research information systems: MAGIC
13
original article
The CRIS can be used and was tested on a wide variety of different static and mobile devices like PC’s, notebooks, smart phones (using Apple iOS and Android), tablet computers and even on E-readers with web browser …. The server-side platforms were a little more limited on Apache HTTP web server and a standard/compatible ANSI SQL relational database (or relational database management system = RDBMS) like MySQL, PostgreSQL, Oracle, MS-SQL… The thorough server-side tests were done only on Linux and MS Windows server platforms using Apache web server and MySQL, PostgreSQL and Oracle databases with only ANSI SQL standard queries. All text data together with interface and (alert) messages are stored in standard Unicode UTF-8 format that is automatically (by using standard HTTP and HTML headers) recognised by the majority of web browsers. Graphical user interface (GUI) is multilingual and all messages are separated from the application, so they can be easily translated to any language. For the time being, there are two languages available: Slovene and English. The MAGIC and generated CRIS CGI scripts are written in Perl scripting language owing to historical reasons and also because Perl is stable, fast and reliable programming language, which runs on vast majority of computer platforms. The execution of Perl CGI scripts with Apache mod_perl module is still one of the most optimal bearing
in mind that major data manipulation is done by a relational database.
CRIS definitions in application definition language (ADL) The CRIS parameters definitions are done together with users interface definitions in a custom made and intuitive English-language-like formalism called application definition language (ADL) [18], which is a mixture of normal text (text for user interface, alerts, messages and visual field names is in chosen local language) and XHTML and SQL languages. For the time being, the definitions can be written in MS Excel data sheets (one variable or title per line) or as a structured (tab delimited) UTF-8 ASCII text files (in an editor like, e.g. Notepad ++). One MS Excel sheet or one text file represents one data entry form or database view (one form can be combined from parameters defined in one or more separate SQL tables with static and repeatable parts of data entryform). There are two special data sheets. One is reserved for end-users (meta) data definitions and other for data entry forms order (how the data entry forms are ordered in a GUI). The example of ADL definition sheets in MS Excel are shown in Fig. 1.
Fig. 1 Example of definition sheets in MS Excel
13
Lightweight application for generating clinical research information systems: MAGIC
3
original article
When the definitions for all data-entry forms (e.g. data model, e.g. Case Report Forms—CRFs and Follow-Up CRFs) are done (as we said that can be done by non-IT professionals!), they are saved as separate ADL text files (using VBA script macro in MS Excel). ADL text files are the necessary input files for the MAGIC. The most important semantic functions of ADL are as follows: ●● description/label (in arbitrary language in UTF-8) and database names (SQL compliant) of data parameters; ●● definition of parameters in SQL data definition language (DDL); ●● definition of links for help files (the help files are written separately in HTML); ●● definition of GUI visual properties (colors, open/close parts of input forms—more detailed visual properties are defined in a separate CSS template); ●● definition of form data input fields in (X)HTML form definitions; ●● validation properties for each parameter/input field (label for obligatory parameters, maximum, minimum and recommended values …); ●● JavaScript mathematical formulas (some data parameters can be dynamically calculated from other parameters and choices); ●● the possibility of including hard-coded code lists; ●● the possibility of including external code lists (or their subsets) as for example SNOMED CT [19], ICDx, LOINC or other (national) code lists like list of healthcare providers ….
Results CRIS application generation First phase The MAGIC parse input ADL text files in more passes where: 1. ADL text files are validated and all known errors and warnings are reported; 2. SQL DDL file is automatically generated and is used for building database tables with required relations regarding the data model definitions in ADL (we call that “applications database”); 3. server-side CGI scripts required for data entry, validation, search, query, edit and export are automatically generated (with all required read-write-execute permissions). The default data search and query engine allows only search by record ID and general search on all data parameters. Data export can be done with selected records in search engine with special BLD text file format and in XML format, which are suitable for direct import (without any parsing or transformations) into standard statistical/analysis tools like R, SPSS, MS Excel, Matlab …;
4. The data validation consistent with validation definitions described in ADL is executed on a client side with the help of special JavaScript library (called vallib6.js). That reduces web server’s load (the data cannot be sent to server if there is any validation error), but for security reasons the same validation (compliant with vallib6.js library) is done on the server’s side after the data is sent to the server. If the execution of JavaScript is disabled on a client, only data validation on server is done; 5. Additional web server configuration files and CRIS application user accounts (and their rights) are generated if necessary. The end-user CRIS application with basic functionalities (data entry, basic search and query, data edit and export) is now generated and working, if the generation was done on operational server. If the generation was not done on a server, the CGI scripts must be transferred to a final operational server destination and the database built by using generated DDL file. The database generation can be done via (web based) database administration tool like phpMyAdmin. With that the first phase is finished and the system is ready-to-use for data-entry, which is the first operational phase needed in every clinical (research) trial. To gain privacy and security, the generated system has already built-in simple user management system through which the user accounts can be divided in groups with different (access) rights. By default for data communication between server and client(s) the secure HTTP (HTTPS) protocol with SSL/TLS encrypted connection is used. Users are authenticated using username and password. If necessary also smartcards with pin can be used (in Slovenia all healthcare professional have a so-called professional health insurance smart card with personal security certificate installed on it). All user’s actions in the system are logged and traceable.
Second phase In second phase, usually the search engine and visual GUI adaptations for a specific user are done. For the second phase, a skilled informatics specialist, programmer or system operator is needed. Some possible second-phase adaptations are as follows: ●● upgrade/change/adapt the search/browse/query/ export engine with additional condition set-up fields; ●● change of visual graphical CSS template (e.g. to add the end-user’s logotype); ●● allow upload and presentation of different (statistical) reports/results that were done elsewhere; ●● adding tools for automatic building of summary reports; ●● adding specific help files (they should be generated as a static HTML files) that were linked in ADL files.
4 Lightweight application for generating clinical research information systems: MAGIC
13
original article Fig. 2 Schematic presentation of clinical research information system (CRIS) generation
Schematic presentation of CRIS generation using MAGIC is shown in Fig. 2.
Use-case example of generated CRIS application Quality of Healthcare Assessment Information System An information system for multi-centric data acquisition and analyses for National Quality of Healthcare Assessment Program (QoHCAP) was designed using the MAGIC. The QoHCAP started in 1999 as a Quality of Healthcare Assessment project jointly supported by major stakeholders in Slovenia like Ministry of Health in Slovenia, Health Insurance Institute of Slovenia, Medical Chamber of Slovenia and University Medical Centre Ljubljana. Within QoHCAP, 40 quality registries for different medical fields were developed with more than 1000 parameters/quality indicators. The information system is used daily by several tenths users from majority of national hospitals and general practitioners. Until now, more than 500,000 assessment forms were entered into the QoHCAP information system. At the beginning, inclusion in the assessment program was voluntary. Since 2011 with the agreement with MoH, the collection and assessment of quality indicators is obligatory for 11 medical specialties. At the same time, a new national web site for publication of analyses and professional and public discussions was setup [20]. Since the beginning, the QoHCAP information system is maintained/generated by MAGIC. Since 2012, the analytical tool for medical content administrators/coordinators was developed. The data collected in the QoHCAP information system can be used for different benchmarking studies [21]. The examples of data entry and analytical GUI are shown in Fig. 3.
Discussion The idea for building an automatic application generation tool started with the first web-based data collection
13
information systems at the end of twentieth century. We persistently searched for other simple generation systems but all proved to be too complex for the requested needs (see aims in introduction). We first developed the so-called AAGIP (Automatic Application Generation in Perl), now renamed to MAGIC, that enables building simple clinical (and other) information systems easier; recent examples are described in [18, 22, 23]. The important issue is also to carefully plan exact data definitions and processes involved in clinical trials’ protocol before the actual clinical trial is started. Many clinical trials and studies would be better or at least optimised regarding the time and effort, if there would be more energy and stress concentrated into the more precise parameter definitions already at the beginning of trial. Of course when the data are collected and analysed, we have to adhere to these definitions. And the electronic information systems with strong validation possibilities can help us a lot. Because of fast changing healthcare environment, we need a very adaptive (clinical research) information system, which would ideally be designed and operated by non-IT professionals usually meaning clinical trial principal investigators (in many cases clinical (healthcare) professionals). By using MAGIC, the “system designers” can be familiar only with normal office applications (MS Excel) and need just very short training time (a few hours maximum, the easiest way is to learn by example) compared with other data management systems like [5–16], which have steep and long learning curve or use complex information system design tools. Generated CRIS also does not need any specific installation on client workstations. Client workstations just need to be “normal internet-equipped” meaning to have internet connection with reasonably new web browser, allowing HTTPS connections and enabled JavaScript (if we do not want to have JavaScript enabled then only server-side validation is active). Other available systems in many cases need installation or enabling of special software modules or client preparation to make the system work [5–16].
Lightweight application for generating clinical research information systems: MAGIC
5
original article
Fig. 3 Example of data entry and analytical graphical user interface in Quality of Healthcare Assessment Information System
Of course our system is not perfect, but it is just good enough to enable fast (and/or optimal and low cost) solution for designing and building CRIS for (multi-centric) data management of small to medium clinical (research) trials. To conclude, we can say that the MAGIC is a very helpful tool for clinical professionals (usually non-IT specialists) that are planning clinical trials. In the future, there could be some improvements like the use of XML schemas (XSD) in HL7 CDA or openEHR archetype formats (also written as XSD) [7, 8, 24–26] as an input parameter/application definition files to MAGIC. The same format could be used (1) for definition files for data interchange, which would enable better interoperability between different generated CRIS applications and other medical information systems and (2) for analyses definition files that would enable performing a (simple, descriptive) automatic statistical analyses and reports from collected data in a database. A statistical analyses module should be developed for that reason. Also adhering the HL7 messaging medical informatics standards [24–26] and Integrating the Healthcare Enterprise (IHE)
profiles [27] would bring some benefits for better CRIS interoperability. The use of MAGIC is not yet open to the general public, but open testing can be arranged by contacting corresponding author. Conflict of Interest The authors declare that there are no actual or potential conflicts of interest in relation to this article.
References 1. Adlassnig KP. Medizinische Informatik in Forschung, Lehre und Patientenversorgung [Medical informatics in research, teaching and patient management]. Wien Klin Wochenschr. 1995;107(1):20–5. 2. Dibacka PL, Bounda Y, Nguema DO, Lell B. Developing information technology at the Medical Research Unit of the Albert Schweitzer Hospital in Lambaréné, Gabon. Wien Klin Wochenschr. 2010;122(Suppl. 1):16–8. 3. Zarin DA, Tse T, Williams RJ, Califf RM, Ide NC. The ClinicalTrials.gov results database—update and key issues. N Engl J Med. 2011;364(9):852–60.
6 Lightweight application for generating clinical research information systems: MAGIC
13
original article 4. O’Reilly EK, Hassell NJ, Snyder DC, Natoli S, Liu I, Rimmler J, Amspacher V, Burnett BK, Parrish AB, Berglund JP, Stacy M. ClinicalTrials.gov reporting: strategies for success at an academic health center. Clin Transl Sci. 2015;8(1):48–51. doi:10.1111/cts.12235. 5. F O Kern E, Beischel S, Stalnaker R, Aron DC, Kirsh SR, Watts SA. Building a diabetes registry from the Veterans Health Administration’s computerized patient record system. J Diabetes Sci Technol. 2008;2(1):7–14. 6. Talend Open Studio (Internet). http://www.talend.com/. Accessed 01 Feb 2015. 7. Garde S, Chen R, Leslie H, Beale T, McNicoll I, Heard S. Archetype-based knowledge management for semantic interoperability of electronic health records. Stud Health Technol Inform. 2009;150:1007–11. 8. Kashfi H, Torgersson O. Supporting openEHR Java desktop application developers. Stud Health Technol Inform. 2011;169:724–8. 9. Mohammed-Rajput NA, Smith DC, Mamlin B, Biondich P, Doebbeling BN; for the Open MRS Collaborative Investigators. OpenMRS, a global medical records system collaborative: factors influencing successful implementation. AMIA Annu Symp Proc. 2011;2011:960–8. 10. Leroux H, McBride S, Gibson S. On selecting a clinical trial management system for large scale, multi-centre, multimodal clinical research study. Stud Health Technol Inform. 2011;168:89–95. 11. Inchingolo P, Beltrame M, Bosazzi P, Cicuta D, Faustini G, Mininel S, et al. O3-DPACS Open-Source Image-Data Manager/Archiver and HDW2 Image-Data Display: an IHEcompliant project pushing the e-health integration in the world. Comput Med Imaging Graph. 2006;30:391–406. 12. IBM Cognos—Clinical Trial Management Software. http:// www-01.ibm.com/software/analytics/cognos/clinicaltrial-management-software.html. Accessed 01 Feb 2015. 13. Symetric Sciences Clinical Data Management Services. http://www.symetric.ca/. Accessed 01 Feb 2015. 14. Thyvalikakath TP, Monaco V, Thambuganipalle HB, Schleyer T. A usability evaluation of four commercial dental computer-based patient record systems. J Am Dent Assoc. 2008;139(12):1632–42.
13
15. Klipin M, Mare I, Hazelhurst S, Kramer B. The process of installing REDCap, a web based database supporting biomedical research: the first year. Appl Clin Inform. 2014;5(4):916–29. doi:10.4338/ACI-2014-06-CR-0054. 16. Schrimpf D, Haag M, Pilz LR. Possible combinations of electronic data capture and randomization systems. Principles and the realization with RANDI2 and OpenClinica. Methods Inf Med. 2014;53(3):202–7. doi:10.3414/ME13-01-0074. 17. Chromatic. Extreme programming pocket guide. Sebastopol: O’Reilly Media; 2003. 18. Dimec J, Leskosek B. Building co-operative bibliographic databases in European bioethics: a contribution from EU New Member States. Comput Methods Programs Biomed. 2007;87(2):123–31. 19. Fung KW, Xu J, Rosenbloom ST, Mohr D, Maram N, Suther T. Testing three problem list terminologies in a simulated data entry environment. AMIA Annu Symp Proc. 2011;2011:445–54. 20. National Quality of Healthcare Assessment Program (in Slovene). http://www.kakovost.si. Accessed 01 Feb 2015. 21. Pal M, Leskošek B, Pajntar M, Ferk P. Evaluation of arterial hypertension control in family practice in Slovenia. Slov Med J. 2014;83(4):299–310. ISSN 1318-0347. 22. Leskošek B. A clinical research information system: two years experience of prospective observational study in oncology. AMIA Annu Symp Proc. 2009 Nov 14; 2009;930. 23. Leskošek B. A clinical research information system: an example of prospective observational study in oncology. AMIA Annu Symp Proc. Nov 6 2008;1021. 24. Rajeev D, Staes C, Evans RS, Price A, Hill M, Mottice S, Risk I, Rolfs R. Evaluation of HL7 v2.5.1 electronic case reports transmitted from a healthcare enterprise to public health. AMIA Annu Symp Proc. 2011;2011:1144–52. 25. Dolin RH, Rogers B, Jaffe C. Health level seven interoperability strategy: big data, incrementally structured. Methods Inf Med. 2015;54(1):75–82. doi:10.3414/ME14-01-0030. 26. Health Level Seven (HL7) International. http://www.hl7. org. Accessed 01 Feb 2015. 27. Integrating the Healthcare Enterprise (IHE). http://www. ihe.net. Accessed 01 Feb 2015.
Lightweight application for generating clinical research information systems: MAGIC
7
Lightweight application for generating clinical research information systems: MAGIC Brane Leskošek · Marjan Pajntar
Received: 22 January 2013 / Accepted: 20 April 2015 © Springer-Verlag Wien 2015
Summary Background Our purpose was to build and test a lightweight solution for generating clinical research information systems (CRIS) that would allow non-IT professionals with basic knowledge of computer usage to quickly define and build a ready-to-use, safe and secure webbased clinical research system for data management. We use the acronym MAGIC (Medical Application Generator InteraCtive) for the system. Methods The generated CRIS should be very easy to build and use, so a common LAMP (Linux Apache MySQL Perl) platform was used, which also enables short development cycles. The application was built and tested using eXtreme Programming (XP) principles by a small development team consisting of one informatics specialist, one physician and one graphical designer/ programmer. Results The parameter and graphical user interface (GUI) definitions for the CRIS can be made by non-IT professionals using an intuitive English-language-like formalism called application definition language (ADL). From these definitions, the MAGIC builds an end-user CRIS that can be used on a wide variety of platforms (from B. Leskošek, PhD, EE () Faculty of Medicine, Institute for Biostatistics and Medical Informatics, University of Ljubljana, Vrazov trg 2, 1000 Ljubljana, Slovenia e-mail: [email protected] B. Leskošek, PhD, EE Faculty of Medicine, University of Maribor, Taborska ulica 8, 2000 Maribor, Slovenia M. Pajntar, PhD, MD Faculty of Medicine, University of Ljubljana, Vrazov trg 2, 1000 Ljubljana, Slovenia
13
standard workstations to hand-held devices). A working example of a national health-care-quality assessment program is presented to illustrate this process. Conclusion The lightweight application for generating CRIS (MAGIC) has proven to be useful for both clinical and analytical users in real working environment. To achieve better performance and interoperability, we are planning to recompile the application using XML schemas (XSD) in HL7 CDA or openEHR archetypes formats used for parameters definition and for data interchange between different information systems. Keywords Information system · Clinical trial · Electronic health record · Clinical research informatics · Interoperability
Introduction Information technology (IT) has become a key resource for research institutions, providing services such as hardware, software and network maintenance, as well as data management services [1, 2]. With the emerging new technologies, the execution of wide range of new different clinical trials becomes possible. New laboratory equipment and new analytical methods makes possible to do many different new examinations and gives us a lot of new data for each patient or sample. A lot of these different clinical trials/studies [3, 4], which are done every day, need (usually electronic) data management (planning, acquisition, validation, quality and analysis). Even though a rapid development of new IT makes data collection more easy, safe, secure and reliable, the connections between (and among) all these data generated by new equipment and methods and between patients/ samples becomes more and more complex. That arises the need for more complex data management systems that will assure structured quality data collection that is
Lightweight application for generating clinical research information systems: MAGIC
1
original article
needed for quality analyses and results. Because of the complexity of data connections and definitions, the data management still takes a lot of time and effort even in cases when all trial parameters are already defined in advance. Sometimes that takes even majority of trial’s time. The main challenge is that the available solutions (either commercial or open source) for data management [5–16] need a professional information specialist/programmer and have very steep and long learning curve, mainly because the IT solutions tries to cover as wide functionalities and options as possible. The needed “usual” use of IT professionals can be also concluded from [2]. However, there is a lack of simple and not so sophisticated clinical research information systems (CRIS) data management tools that would enable the basic functionality of safe, secure and reliable quality data management and close the gap between simple and complex data management systems. That is especially evident in smaller clinical trials where usually the professional information specialist is not available or it would be too expensive to have it in a research team. But professional information specialist is actually needed for using more complex solutions [12–16]. This situation many times results in poor research (data and quality management) information systems that leads to bad data quality, (long) time delays, unreliable analyses, large expenses, hard-to-adapt IT systems, etc. The purpose of this investigation was to build and test a specific lightweight solution for generation of CRIS (MAGIC—Medical Application Generator InteraCtive). The aims of the MAGIC are as follows: ●● to enable rapid parameters/data definitions by nonIT professionals, which have some very basic knowledge about computer usage and are usually principal medical investigators such as medical specialists— physicians, pharmacologists, pharmacists, molecular biologists …; ●● that only very widespread and basic computer tools are needed for parameters/data and user interface definitions, e.g. MS Excel, any (Unicode UTF-8 enabled) ASCII editor (e.g. Notepad, Notepad ++ …); ●● to get a ready-to-use safe and secure CRIS for fast data management and with low (establishment and maintenance) costs; ●● to have a very intuitive generated CRIS application usage where none or only very short training time is needed for end-users; ●● that no installation or changes or adaptations of endusers’ workstations/computers is needed. The generated CRIS with basic functionalities should be used from as many different computers/operation systems as possible and should run within existing organizational information systems without the need for any changes, adoption or installation of workstations software;
●● that the usage from different dislocated units (e.g. laboratories, institutions …) should be integrated as the basic functionality in the generated CRIS; ●● that the generated CRIS can be easily upgraded with more sophisticated functionalities and different graphical design even though for that a trained and skilled information specialist/programmer is needed.
Methods The CRIS generated using MAGIC should be very easy to build and use and maintained using short development cycles. The primary principle of building method for CRIS using MAGIC was eXtreme Programming (XP) [17]. We use that principle mainly for the end-users application’s (“generated CRIS”) parameters definitions, installation and (intuitive) usage. CRIS application is web based because of the world’s internet penetration with well-tested security mechanisms (like SSL cyphering algorithms …). All communication and data transfer is done only via http(s) protocols. To follow previously mentioned lightweight requests and always present need for fast development cycles, a very common open source LAMP (Linux Apache MySQL Perl) platform was used. The generation tool was built and tested using XP principles [17] in very small development team consisting from one informatics specialist, one physician and one graphical designer/programmer. To allow usage on a variety of different platforms and from many different places regardless of the internet connection bandwidth, the end-user’s CRIS application should exploit only basic web technologies.
Platforms used for generation tool and CRIS CRIS application exploits only basic web technologies like (X)HTML, JavaScript, HTML CGI, POST and GET methods and cascading style sheets (CSS) for (X)HTML documents’ visual presentation/graphical design. Clinical trials functionality is at least in a (first) phase when data are collected, much more important than application’s graphical design, especially because graphical design could limit the spread of (CRIS) application usage. “Fancy” technologies like Adobe Flash, Silverlight and similar were avoided because they usually adds only more “appealing” (also depending on user’s taste) visual appearance but no or just very limited added functionality. And last but not least important goal was also to develop a robust and operating system-independent solution. The supported and tested client-side platforms are UNIX-like (Linux, BSD, Mac OS X …), MS Windows, iOS, Android, VMS … with many different XHTML compatible browsers like (newer versions of ) Mozilla Firefox, MS Internet Explorer, Google Chrome, Apple Safari, Opera ….
2 Lightweight application for generating clinical research information systems: MAGIC
13
original article
The CRIS can be used and was tested on a wide variety of different static and mobile devices like PC’s, notebooks, smart phones (using Apple iOS and Android), tablet computers and even on E-readers with web browser …. The server-side platforms were a little more limited on Apache HTTP web server and a standard/compatible ANSI SQL relational database (or relational database management system = RDBMS) like MySQL, PostgreSQL, Oracle, MS-SQL… The thorough server-side tests were done only on Linux and MS Windows server platforms using Apache web server and MySQL, PostgreSQL and Oracle databases with only ANSI SQL standard queries. All text data together with interface and (alert) messages are stored in standard Unicode UTF-8 format that is automatically (by using standard HTTP and HTML headers) recognised by the majority of web browsers. Graphical user interface (GUI) is multilingual and all messages are separated from the application, so they can be easily translated to any language. For the time being, there are two languages available: Slovene and English. The MAGIC and generated CRIS CGI scripts are written in Perl scripting language owing to historical reasons and also because Perl is stable, fast and reliable programming language, which runs on vast majority of computer platforms. The execution of Perl CGI scripts with Apache mod_perl module is still one of the most optimal bearing
in mind that major data manipulation is done by a relational database.
CRIS definitions in application definition language (ADL) The CRIS parameters definitions are done together with users interface definitions in a custom made and intuitive English-language-like formalism called application definition language (ADL) [18], which is a mixture of normal text (text for user interface, alerts, messages and visual field names is in chosen local language) and XHTML and SQL languages. For the time being, the definitions can be written in MS Excel data sheets (one variable or title per line) or as a structured (tab delimited) UTF-8 ASCII text files (in an editor like, e.g. Notepad ++). One MS Excel sheet or one text file represents one data entry form or database view (one form can be combined from parameters defined in one or more separate SQL tables with static and repeatable parts of data entryform). There are two special data sheets. One is reserved for end-users (meta) data definitions and other for data entry forms order (how the data entry forms are ordered in a GUI). The example of ADL definition sheets in MS Excel are shown in Fig. 1.
Fig. 1 Example of definition sheets in MS Excel
13
Lightweight application for generating clinical research information systems: MAGIC
3
original article
When the definitions for all data-entry forms (e.g. data model, e.g. Case Report Forms—CRFs and Follow-Up CRFs) are done (as we said that can be done by non-IT professionals!), they are saved as separate ADL text files (using VBA script macro in MS Excel). ADL text files are the necessary input files for the MAGIC. The most important semantic functions of ADL are as follows: ●● description/label (in arbitrary language in UTF-8) and database names (SQL compliant) of data parameters; ●● definition of parameters in SQL data definition language (DDL); ●● definition of links for help files (the help files are written separately in HTML); ●● definition of GUI visual properties (colors, open/close parts of input forms—more detailed visual properties are defined in a separate CSS template); ●● definition of form data input fields in (X)HTML form definitions; ●● validation properties for each parameter/input field (label for obligatory parameters, maximum, minimum and recommended values …); ●● JavaScript mathematical formulas (some data parameters can be dynamically calculated from other parameters and choices); ●● the possibility of including hard-coded code lists; ●● the possibility of including external code lists (or their subsets) as for example SNOMED CT [19], ICDx, LOINC or other (national) code lists like list of healthcare providers ….
Results CRIS application generation First phase The MAGIC parse input ADL text files in more passes where: 1. ADL text files are validated and all known errors and warnings are reported; 2. SQL DDL file is automatically generated and is used for building database tables with required relations regarding the data model definitions in ADL (we call that “applications database”); 3. server-side CGI scripts required for data entry, validation, search, query, edit and export are automatically generated (with all required read-write-execute permissions). The default data search and query engine allows only search by record ID and general search on all data parameters. Data export can be done with selected records in search engine with special BLD text file format and in XML format, which are suitable for direct import (without any parsing or transformations) into standard statistical/analysis tools like R, SPSS, MS Excel, Matlab …;
4. The data validation consistent with validation definitions described in ADL is executed on a client side with the help of special JavaScript library (called vallib6.js). That reduces web server’s load (the data cannot be sent to server if there is any validation error), but for security reasons the same validation (compliant with vallib6.js library) is done on the server’s side after the data is sent to the server. If the execution of JavaScript is disabled on a client, only data validation on server is done; 5. Additional web server configuration files and CRIS application user accounts (and their rights) are generated if necessary. The end-user CRIS application with basic functionalities (data entry, basic search and query, data edit and export) is now generated and working, if the generation was done on operational server. If the generation was not done on a server, the CGI scripts must be transferred to a final operational server destination and the database built by using generated DDL file. The database generation can be done via (web based) database administration tool like phpMyAdmin. With that the first phase is finished and the system is ready-to-use for data-entry, which is the first operational phase needed in every clinical (research) trial. To gain privacy and security, the generated system has already built-in simple user management system through which the user accounts can be divided in groups with different (access) rights. By default for data communication between server and client(s) the secure HTTP (HTTPS) protocol with SSL/TLS encrypted connection is used. Users are authenticated using username and password. If necessary also smartcards with pin can be used (in Slovenia all healthcare professional have a so-called professional health insurance smart card with personal security certificate installed on it). All user’s actions in the system are logged and traceable.
Second phase In second phase, usually the search engine and visual GUI adaptations for a specific user are done. For the second phase, a skilled informatics specialist, programmer or system operator is needed. Some possible second-phase adaptations are as follows: ●● upgrade/change/adapt the search/browse/query/ export engine with additional condition set-up fields; ●● change of visual graphical CSS template (e.g. to add the end-user’s logotype); ●● allow upload and presentation of different (statistical) reports/results that were done elsewhere; ●● adding tools for automatic building of summary reports; ●● adding specific help files (they should be generated as a static HTML files) that were linked in ADL files.
4 Lightweight application for generating clinical research information systems: MAGIC
13
original article Fig. 2 Schematic presentation of clinical research information system (CRIS) generation
Schematic presentation of CRIS generation using MAGIC is shown in Fig. 2.
Use-case example of generated CRIS application Quality of Healthcare Assessment Information System An information system for multi-centric data acquisition and analyses for National Quality of Healthcare Assessment Program (QoHCAP) was designed using the MAGIC. The QoHCAP started in 1999 as a Quality of Healthcare Assessment project jointly supported by major stakeholders in Slovenia like Ministry of Health in Slovenia, Health Insurance Institute of Slovenia, Medical Chamber of Slovenia and University Medical Centre Ljubljana. Within QoHCAP, 40 quality registries for different medical fields were developed with more than 1000 parameters/quality indicators. The information system is used daily by several tenths users from majority of national hospitals and general practitioners. Until now, more than 500,000 assessment forms were entered into the QoHCAP information system. At the beginning, inclusion in the assessment program was voluntary. Since 2011 with the agreement with MoH, the collection and assessment of quality indicators is obligatory for 11 medical specialties. At the same time, a new national web site for publication of analyses and professional and public discussions was setup [20]. Since the beginning, the QoHCAP information system is maintained/generated by MAGIC. Since 2012, the analytical tool for medical content administrators/coordinators was developed. The data collected in the QoHCAP information system can be used for different benchmarking studies [21]. The examples of data entry and analytical GUI are shown in Fig. 3.
Discussion The idea for building an automatic application generation tool started with the first web-based data collection
13
information systems at the end of twentieth century. We persistently searched for other simple generation systems but all proved to be too complex for the requested needs (see aims in introduction). We first developed the so-called AAGIP (Automatic Application Generation in Perl), now renamed to MAGIC, that enables building simple clinical (and other) information systems easier; recent examples are described in [18, 22, 23]. The important issue is also to carefully plan exact data definitions and processes involved in clinical trials’ protocol before the actual clinical trial is started. Many clinical trials and studies would be better or at least optimised regarding the time and effort, if there would be more energy and stress concentrated into the more precise parameter definitions already at the beginning of trial. Of course when the data are collected and analysed, we have to adhere to these definitions. And the electronic information systems with strong validation possibilities can help us a lot. Because of fast changing healthcare environment, we need a very adaptive (clinical research) information system, which would ideally be designed and operated by non-IT professionals usually meaning clinical trial principal investigators (in many cases clinical (healthcare) professionals). By using MAGIC, the “system designers” can be familiar only with normal office applications (MS Excel) and need just very short training time (a few hours maximum, the easiest way is to learn by example) compared with other data management systems like [5–16], which have steep and long learning curve or use complex information system design tools. Generated CRIS also does not need any specific installation on client workstations. Client workstations just need to be “normal internet-equipped” meaning to have internet connection with reasonably new web browser, allowing HTTPS connections and enabled JavaScript (if we do not want to have JavaScript enabled then only server-side validation is active). Other available systems in many cases need installation or enabling of special software modules or client preparation to make the system work [5–16].
Lightweight application for generating clinical research information systems: MAGIC
5
original article
Fig. 3 Example of data entry and analytical graphical user interface in Quality of Healthcare Assessment Information System
Of course our system is not perfect, but it is just good enough to enable fast (and/or optimal and low cost) solution for designing and building CRIS for (multi-centric) data management of small to medium clinical (research) trials. To conclude, we can say that the MAGIC is a very helpful tool for clinical professionals (usually non-IT specialists) that are planning clinical trials. In the future, there could be some improvements like the use of XML schemas (XSD) in HL7 CDA or openEHR archetype formats (also written as XSD) [7, 8, 24–26] as an input parameter/application definition files to MAGIC. The same format could be used (1) for definition files for data interchange, which would enable better interoperability between different generated CRIS applications and other medical information systems and (2) for analyses definition files that would enable performing a (simple, descriptive) automatic statistical analyses and reports from collected data in a database. A statistical analyses module should be developed for that reason. Also adhering the HL7 messaging medical informatics standards [24–26] and Integrating the Healthcare Enterprise (IHE)
profiles [27] would bring some benefits for better CRIS interoperability. The use of MAGIC is not yet open to the general public, but open testing can be arranged by contacting corresponding author. Conflict of Interest The authors declare that there are no actual or potential conflicts of interest in relation to this article.
References 1. Adlassnig KP. Medizinische Informatik in Forschung, Lehre und Patientenversorgung [Medical informatics in research, teaching and patient management]. Wien Klin Wochenschr. 1995;107(1):20–5. 2. Dibacka PL, Bounda Y, Nguema DO, Lell B. Developing information technology at the Medical Research Unit of the Albert Schweitzer Hospital in Lambaréné, Gabon. Wien Klin Wochenschr. 2010;122(Suppl. 1):16–8. 3. Zarin DA, Tse T, Williams RJ, Califf RM, Ide NC. The ClinicalTrials.gov results database—update and key issues. N Engl J Med. 2011;364(9):852–60.
6 Lightweight application for generating clinical research information systems: MAGIC
13
original article 4. O’Reilly EK, Hassell NJ, Snyder DC, Natoli S, Liu I, Rimmler J, Amspacher V, Burnett BK, Parrish AB, Berglund JP, Stacy M. ClinicalTrials.gov reporting: strategies for success at an academic health center. Clin Transl Sci. 2015;8(1):48–51. doi:10.1111/cts.12235. 5. F O Kern E, Beischel S, Stalnaker R, Aron DC, Kirsh SR, Watts SA. Building a diabetes registry from the Veterans Health Administration’s computerized patient record system. J Diabetes Sci Technol. 2008;2(1):7–14. 6. Talend Open Studio (Internet). http://www.talend.com/. Accessed 01 Feb 2015. 7. Garde S, Chen R, Leslie H, Beale T, McNicoll I, Heard S. Archetype-based knowledge management for semantic interoperability of electronic health records. Stud Health Technol Inform. 2009;150:1007–11. 8. Kashfi H, Torgersson O. Supporting openEHR Java desktop application developers. Stud Health Technol Inform. 2011;169:724–8. 9. Mohammed-Rajput NA, Smith DC, Mamlin B, Biondich P, Doebbeling BN; for the Open MRS Collaborative Investigators. OpenMRS, a global medical records system collaborative: factors influencing successful implementation. AMIA Annu Symp Proc. 2011;2011:960–8. 10. Leroux H, McBride S, Gibson S. On selecting a clinical trial management system for large scale, multi-centre, multimodal clinical research study. Stud Health Technol Inform. 2011;168:89–95. 11. Inchingolo P, Beltrame M, Bosazzi P, Cicuta D, Faustini G, Mininel S, et al. O3-DPACS Open-Source Image-Data Manager/Archiver and HDW2 Image-Data Display: an IHEcompliant project pushing the e-health integration in the world. Comput Med Imaging Graph. 2006;30:391–406. 12. IBM Cognos—Clinical Trial Management Software. http:// www-01.ibm.com/software/analytics/cognos/clinicaltrial-management-software.html. Accessed 01 Feb 2015. 13. Symetric Sciences Clinical Data Management Services. http://www.symetric.ca/. Accessed 01 Feb 2015. 14. Thyvalikakath TP, Monaco V, Thambuganipalle HB, Schleyer T. A usability evaluation of four commercial dental computer-based patient record systems. J Am Dent Assoc. 2008;139(12):1632–42.
13
15. Klipin M, Mare I, Hazelhurst S, Kramer B. The process of installing REDCap, a web based database supporting biomedical research: the first year. Appl Clin Inform. 2014;5(4):916–29. doi:10.4338/ACI-2014-06-CR-0054. 16. Schrimpf D, Haag M, Pilz LR. Possible combinations of electronic data capture and randomization systems. Principles and the realization with RANDI2 and OpenClinica. Methods Inf Med. 2014;53(3):202–7. doi:10.3414/ME13-01-0074. 17. Chromatic. Extreme programming pocket guide. Sebastopol: O’Reilly Media; 2003. 18. Dimec J, Leskosek B. Building co-operative bibliographic databases in European bioethics: a contribution from EU New Member States. Comput Methods Programs Biomed. 2007;87(2):123–31. 19. Fung KW, Xu J, Rosenbloom ST, Mohr D, Maram N, Suther T. Testing three problem list terminologies in a simulated data entry environment. AMIA Annu Symp Proc. 2011;2011:445–54. 20. National Quality of Healthcare Assessment Program (in Slovene). http://www.kakovost.si. Accessed 01 Feb 2015. 21. Pal M, Leskošek B, Pajntar M, Ferk P. Evaluation of arterial hypertension control in family practice in Slovenia. Slov Med J. 2014;83(4):299–310. ISSN 1318-0347. 22. Leskošek B. A clinical research information system: two years experience of prospective observational study in oncology. AMIA Annu Symp Proc. 2009 Nov 14; 2009;930. 23. Leskošek B. A clinical research information system: an example of prospective observational study in oncology. AMIA Annu Symp Proc. Nov 6 2008;1021. 24. Rajeev D, Staes C, Evans RS, Price A, Hill M, Mottice S, Risk I, Rolfs R. Evaluation of HL7 v2.5.1 electronic case reports transmitted from a healthcare enterprise to public health. AMIA Annu Symp Proc. 2011;2011:1144–52. 25. Dolin RH, Rogers B, Jaffe C. Health level seven interoperability strategy: big data, incrementally structured. Methods Inf Med. 2015;54(1):75–82. doi:10.3414/ME14-01-0030. 26. Health Level Seven (HL7) International. http://www.hl7. org. Accessed 01 Feb 2015. 27. Integrating the Healthcare Enterprise (IHE). http://www. ihe.net. Accessed 01 Feb 2015.
Lightweight application for generating clinical research information systems: MAGIC
7
