Software engineering and implementation

The development of the eSMAF followed a modified waterfall model with feedback loops from users. The modified waterfall model is a software development model in which development is seen as flowing steadily downwards (like a waterfall) through the phases of requirements analysis and functional specifications, design and implementation of software architecture, software testing and validation, and software deployment and maintenance [22]. Each phase has a well-defined starting and ending point with a closed feedback loop with identifiable deliverables to the next phase. The eSMAF project started in 2003 and culminated in the last year with the commercial deployment of the software in several organizations and institutions. The timeline for the development phases of the eSMAF is illustrated in Figure 1. The details of the development, implementation and deployment of the software are described below.

Design and implementation of software architecture

After taking into consideration the complexities of the information flow of the SMAF data (from data entry to consultation to report and data extraction by clinical personnel, managers and researchers) and its variations depending on the context of use (homecare, hospital care or research), very early in the project we abandoned the idea of embedding the eSMAF in existing proprietary clinical information systems currently in use. In order to transmit information from this stand-alone system back to the patient's electronic medical record or other databases, we chose static merging of individual or aggregated group data from the eSMAF database as CVS files using an export function. An overview of the software architecture for the eSMAF is summarized in Figure 2. In keeping with the initial requirements list and projected use of the software, we chose as the basis for the software architecture 1) Microsoft Access as a back end (database) to store information, and 2) a Visual Basic client to act as a front end (software interface) for easy data insertion, retrieval and reporting. The format of the database was selected according to projected deployment constraints (limited availability of on-site and remote technological support and cost issues) and ease of installation (no server required).

The software architecture is based on five overlapping hierarchical levels (group, subgroup, client, client test, test detail). Each level provides a different view of the database (ex: global aggregated results at the group level to detailed individual information at the test detail level) from which specific functions such as user and workflow management, client record management, online help, graphs, reports, data extraction or printing can be executed. The core of the software architecture is the "Clients" level where a unique subject identifier is created, tests or data entry masks are linked to this unique identifier (Client Test level) and data entry functions (Test edition) are applied. The unique identifier at the "Clients" level can be associated with "Groups" or "Subgroups" to aggregate clients' records in user-defined categories, thus facilitating navigation and viewing of records in the database. For example, a group could be used to represent a health care center with the subgroups representing the different units in this center. In this way, the user can view individual patient records for a subgroup of clients in an institution. The group and subgroup can also be used to limit access to records to a specific category of user.

Software functions of eSMAF

The principal functions of the eSMAF software are summarized below and include: (1) Navigation interface and records management utility; (2) Users and workflow management; (3) Client record module and data entry; (4) Report function and care planning; (5) Search functions; and (6) Data extraction function and online help.

Navigation interface and records management utility

A standard interface is used throughout the hierarchical levels of the software. The layout of the interface is shown in Figure 3. There are three main work areas: the main area, the toolbar area and the navigation toolbar area. The main area is used to display or edit information. The toolbar area provides access to the software functions, depending on the hierarchical level the user is on. At the group level, functions include data management functions, report and search functions, and the software management control panel. The interface at the group and client level consists of buttons and list views which can be used to display patients or tests in different modes: icons, list and detailed view. The detailed view includes a sorting option that allows users to display records according to specific criteria (ex: date of creation or Iso-SMAF profile). The navigation toolbar allows users to navigate across hierarchical levels of the software.

Users and workflow management

Workflow management options (administrator and user privileges, records management, etc.) are set through a control panel accessible from the group level interface. Figure 4 shows the interface of this control panel and the principal workflow management options. Access privileges for users are defined according to user groups created by the administrator users. Each user is associated with one user group. The user group determines what kind of access the user in question has (logging into the database, accessing groups or subgroups in the client records, viewing and/or editing records or portions of records). A "Utilities/Options" panel accessible only to users with administrative privileges is used to configure options for the software (ex: path to the location of the database). Traceability features (ex: what was changed in a given record and when) can be viewed through a software log tab as all user actions are logged in the database.

Client record module and data entry

The client record module consists of a data repository of evaluations completed for a given client, allowing users with administrative privileges to generate new evaluation masks. Each evaluation can then be filed and edited. The main evaluation used in the software is the SMAF test, but it is also possible to fill out an information record containing nominal information, a clinical note and a care table (see Report function and care planning below). The SMAF test interface is shown in Figure 5. A test can be locked and unlocked, depending on the access the user has. Data are entered by accessing questions using the overall questionnaire view or using the navigation functions in the question view. Complementary information for a given item can be entered through tab navigation in the question view.

Report function and care planning

In order to continue support for paper-based clinical pathways, a report module was incorporated to allow users to generate paper summary reports or detailed reports on any given client or group of clients. Reports can take several forms and include: individualized summary, distribution graphs on specific indices (ex: Iso-SMAF profiles) of selected groups of patients, line graphs charting the evolution of a specific category of data for an individual across multiple evaluations, and bar charts showing specific changes between the two last evaluations of a given individual. Report functions can be applied at the client level or aggregate clients level by manually selecting them in the navigation interface and records management utilities or through selected outputs from the search engine. The principal report function outputs are illustrated in Figure 6.

In addition to these reporting features, users can automatically generate, for a given client or clients, a color-coded graphical care table that, for each section of the SMAF, summarizes the level of assistance needed by this or these individuals. The care table can be generated individually using data entered on the SMAF scale and printed. An illustration of a care table is presented in Figure 7. For a specific sphere of activities (ex: ADL), colors represent the level of assistance needed by the client for a given task (ex: bathing). The printed care table is then displayed in the client's room and serves as a visual reference to help guide the care provided to the client by nursing staff and orderlies.

Search functions

Search functions were implemented at the aggregate client level to allow users to search and parse information from the database according to user-defined parameters. The search functions were implemented so that it would be possible to generate user-defined queries on all levels of the database (tests, patient information, groups, etc.) and execute report and data extraction functions on a subset of specific records. The interface for the search functions is illustrated in Figure 8. To do a search, search parameters need to be specified. There are no limits to the number of parameters that can be specified. Individual parameters are specified for a category of information (ex: score on a given section or question of the SMAF). Strings of parameters can be grouped together with the logical operator "And" or "Or" and saved as a specific search that can be reapplied at any time. Upon execution of a search, records (clients) matching the search parameters are clustered together as a group of records. Report functions and data extraction functions on this group of records or on individual records can then be applied.

Data extraction function and online help

Individual or aggregated group data can be transferred and merged with other datasets using the export function. Selected records or sections of records within the database can be exported as CVS files. The CVS file can be imported to another database, spreadsheet files and statistical applications such as SPSS. The data extraction function is illustrated in Figure 9. Online help in the eSMAF can be consulted to address issues with the software itself and to crosscheck scoring procedures on the SMAF evaluation. The software help system can be accessed from anywhere in the software. The SMAF help is only available when editing a SMAF test, and provides information on how to complete a specific question on the test. The help function is illustrated in Figure 10.

Intended use of the software

The eSMAF's principal intended use is to streamline the collection of SMAF and Iso-SMAF data in clinical settings while offering a way to control the source and quality of the data entered, facilitate the aggregation, viewing and extraction of these data according to user-defined needs, and empower users to easily merge this information with other databases or data sources. For example, in the context of program evaluation research, the eSMAF was successfully deployed to collect SMAF outcome measures periodically over a 2-year period at 19 long-term public home care institutions in an administrative region of Quebec, Canada [24]. Medical archives personnel (n = 57 users) were trained on the use of the software prior to the start of the project. A key individual was identified in each facility to manage user questions and technical problems. This person was in contact with a district coordinator, who was the contact point for the research team. Clinical assessments in the eSMAF database from each institution were merged with administrative databases in order to objectively describe the relationship between the needs of older adults as established from Iso-SMAF profiles and the actual services provided to them by home care programs. SMAF data from 8000 patient records over a 2-year period were collected. As an example of the accessibility of these data, retrospective analysis of the data clearly illustrated the presence of a gap between the levels of home care services provided and the needs of disabled individuals, thus providing both researchers and clinical managers with arguments for changes in budget allocation for services in this administrative region. The data continue to be collected and used by administrative personnel to benchmark and document the autonomy of the client population and the services offered.

In this deployment, some barriers to implementation were encountered. The first was related to the use of small monitors and low resolution screens by some of the users. At the time many of the users were working with a screen resolution of 800 × 600 on 15-inch monitors. As the software interface was designed to work with a screen resolution of at least 1024 × 768, this caused a major problem. This problem was resolved by making the software automatically switch to the required resolution in the operating system (i.e. Windows) when launching, then switching back when it was closed. Another issue arose when the eSMAF was installed in many different network configurations in different physical locations, and regular updates were required because of its development stage. While the IT support needed to install and maintain the software in a server environment is minimal, this support was not readily available at most sites as IT support was contracted out. This introduced delays in granting user administrative and access privileges to the database. Communications with each site's IT support department was also difficult. When an update to the software was needed, it was not installed at the same time at each location. This caused problems when trying to find if a reported bug was related to a new or an old version and required close follow-up of installed versions in each physical location. Once installed and configured, the eSMAF performed as expected.

For clinical purposes, it is also possible to incorporate and analyze SMAF data and Iso-SMAF profiles on a regular basis, thereby rendering them accessible for decision-making in day-to-day operations (define eligibility criteria, support planning and evaluation processes, organize resource allocations) at different clinical and administrative levels. For example, from patient SMAF evaluations using the search function, department heads or hospital unit managers can estimate the clinical work load associated with a number of beds in a unit by looking at specific functional autonomy indicators (incontinence, dressing and eating) of patients in this unit. Care plans can also be circulated more freely and updated more easily. Prior to the development of the eSMAF, the clinical staff generated care tables for each client manually by putting color-coded stickers on each section of the care table and they had to crosscheck that they corresponded with the current SMAF data on record. The eSMAF's care table function offers an automated means to efficiently generate care tables for large numbers of clients while insuring that the outputs in these care tables correspond to the clients' on-record functional status.

Planned future developments

Using feedback from current users, we are looking at optimizing the data management features of the eSMAF to facilitate its integration in institutional clinical workflow and/or its use in a health services research context. While the software in its current version has been used successfully in the context of multiple users (one database stored on a local area network with multiple controlled access to the database), the underlying database infrastructure is not optimized for large-scale networked use (i.e. database with more than 5000 records with simultaneous access from more than 40 users from multiple geographically separate sites). Initial user needs assessment and budgetary constraints governed the choice of the database infrastructure to a readily available non-commercial grade database format. The existing database platform (Microsoft Access 2000) is not robust enough to operate under such conditions. Therefore, to expand the usability of the software under this context of use, a Web-based eSMAF supported with a commercial grade database format (ex: My SQL) is being considered.

Dynamic integration of the eSMAF software with existing clinical information systems will also a play a critical role in its successful deployment in large-scale hospital settings. This integration is possible with the existing software architecture but depends to a large extent on establishing communication between systems and establishing database synchronization rules that fit the workflow and needs of the application. Depending on user interest, another development area will be to offer other language versions. As the SMAF instrument has been translated into 7 languages, the eSMAF could be ported to other languages. The underlying programming architecture related to the operation and use of the software is not language-specific. Indeed the programming architecture is such that most of the user interface shell (data entry mask of the SMAF) can be customized rapidly while still maintaining the current database structure using tables containing content in any language.