General Design/Implementation Decisions

To maximize the chances of user adoption it was decided that the PMIS must mimic existing processes and workflows so as to be as minimally invasive as possible. Furthermore, the PMIS should attempt to replicate the metaphor of the familiar paper documents and other forms traditionally used in healthcare delivery. Given the unfamiliarity of the target user group with computer technology and particularly their lack of skill in typing, two design decisions were made regarding the user interface that differentiated the solution from information system implementations traditionally seen in healthcare settings. The first decision related to screen design and the method of data entry. As computer displays and screen resolutions have increased in size system developers have had a tendency to increase the amount of information contained on a single screen. This often results in overly complex screens, which can be overwhelming and harder to navigate for the user.

A touchscreen-based graphical user interface would be used for human computer interaction. Button size and inter-button spacing would be optimized for ease of use with a finger rather than a mouse.

Figure 3: Sample screen shot representing a prototypical layout of controls

User interfaces should be designed to match the skills, experience and expectations of the anticipated users. The decision was made to use a “wizard-like” approach to capturing

information. Rather than having multiple data entry fields on a single screen, each screen would be dedicated to collecting a single piece of data.

Generally speaking screens would be designed with the upper portion of the screen displaying the set of data elements being captured for the particular task at hand and the lower portion of the screen providing data entry options for each data element. In the sample screen shown in Figure 3, the upper portion shows the information being collected to register a patient. In the lower portion a hierarchically organized list of geographical locations is shown allowing the user to complete the data field for the indicating the location where the patient currently resides. This motif would be used throughout the application for consistency.

The second design decision related to the method of human-computer interaction. It was decided that a touchscreen interface would provide the optimal solution in usability. While touchscreen technology had been around for more than 15 years, it has only been adopted in limited vertical applications. Touchscreens lend themselves to applications more focused on information retrieval than information gathering. Using a touchscreen in place of a keyboard greatly limits the capacity to enter free text. However, considering the focus on collecting categorical data the touchscreen seemed to be a perfect choice. A small amount of free text would be necessary, for example entering patient names. To accommodate for this a virtual keyboard would be displayed on the screen.

To meet the need for high system accessibility multiple workstations would need to be deployed. It was decided that three workstations would be deployed in the outpatient area, one at the patient registration area adjacent to the outpatient clinic, one in the pediatric medical records office and one at each of the three nursing stations on the pediatric wards, for a total of eight workstations.

To accommodate the necessity for mobility in the outpatient area wireless networking would be employed. A wireless access point was installed in the outpatient area.

To protect against power failures that would prevent workstations from functioning, a large (96 Amp Hour) deep-cycle battery would be connected to each workstation. The capacity of the battery would need to be sufficiently large to enable workstations to run continuously for more than 12 hours from a single charge. Both the main server and the network electronics would need to be backed-up with an uninterruptible power supply (UPS). This backup

arrangement was not required to run the PMIS for long periods of time, just sufficiently long for the hospital generator to come online.

A client-server architecture was chosen. A failure of the server would cause the entire system to fail. Consequently measures would need to be taken to maximize the reliability and availability of the server.

Theft of equipment was considered to be a serious threat to the success of the PMIS. To protect against theft, the server would need to be located in a highly secure room. The medical records office would also need to be secured. Given that workstations would be more accessible they would be at greater risk of theft. Increasing the security of workstations by locking them in rooms away from patients confounded the goal of having workstations used in realtime at the point-of-care. It was decided that no data would reside on the workstations to eliminate the risk of loss of data should a workstation be stolen.

While the PMIS emphasized the collection of categorical data over free text, patient names could not reasonably be picked from a list and would need to be entered using some form of keyboard. An onscreen keyboard would be developed for this. Given the high frequency of particularly common Malawian names, particularly surnames, a list box containing approximately 1,000 common names was displayed onscreen to allow for auto-completion purpose. Given the low levels of computer literacy among the target users there seemed to be little advantage to using the QWERTY format for the keyboard. Consequently an alphabetic layout was adopted.

The barcode 39 (3-of-9) format was chosen for use on the adhesive labels as it can represent both alphabetic and numeric characters and is able to be read by the vast majority of barcode scanners on the (used) market.