Research design

This study was conducted in two phases carried out in Ontario, Canada. In Phase One, database requirements were gathered from consultations with an expert panel and through a review of the CP@clinic protocol, algorithms and assessment tools to determine the database platform, functionality, contents, and security required to meet the needs of CP@clinic. In Phase Two, utility testing was conducted with potential database users to ensure a user-friendly interface. A cyclical iterative design process was utilized where knowledge gained from one iteration was used to inform the next iteration [7,8]. Since it was not possible to know upfront what the final database should look like, an approach was used that allowed multiple product refinements. The study was approved by the Hamilton Integrated Research Ethics Board (HiREB).

Participants

Expert Panel: The CP@clinic research team participated in this study as an expert panel. The multidisciplinary research team consists of a family doctor with over 30 years of experience using electronic medical records, a public health nurse with over 30 years of experience using medical records, a paramedic with over 15 years of experience with record keeping for acute emergency medical services, a research associate with over 25 years of experience in database and statistical programs, a program coordinator, and an Information Technology (IT) manager with over 20 years of experience maintaining and hosting electronic medical records and networking.

Potential Database Users: Although the current CP@clinic model had paramedics staffing the program, there was potential for a volunteer peer-led, paramedic-supervised model in the future; therefore, it was important to consider older adults (55 years and older) as potential users when designing the database. A purposive sample of five paramedics and ten older adults who shared similar characteristics with potential volunteers were recruited to test the database usability. All participants were over the age of 18. We recruited 5 paramedics and 10 older adults to allow us to conduct the iterative process of developing, revising, and improving the database. Three paramedics participated in the initial development of the database, 5 older adults participated in the initial testing, 5 older adults participated in repeat testing after revisions were incorporated; 2 of the paramedics involved in the initial development and 2 new paramedics were involved in the final testing.

Data gathering and testing procedures

To efficiently implement CP@clinic, a project-specific smart database was required with the capability to:

• Capture initial and follow-up data
• Guide the paramedic personnel in administering the risk assessments survey
• Calculate and summarize the participants’ risk factor scores
• Direct individuals in different predetermined clinical algorithm pathways according to their risk levels and evidence-based medical directives
• Provide information about local resources available to manage risk factors that are tailored to the participants’ situation
• Provide automated alerts based on specified evidence-based algorithms to facilitate ongoing data capture (e.g. 6 month follow-up for fasting blood glucose)
• Generate automated reports for researchers, participants, and family physicians
• Securely sync data from project sites to a central, project-specific server

In Phase One, the expert panel reviewed the contents, platform, functionality, and data security elements required for the database. Examining the CP@clinic protocol, the database should include participant demographic information, physician contact information (including a stored physician list), physical measurements (e.g. height, fasting blood glucose), medical and medication history, a resource list, and a record of risk factor discussions. It also needed to include publicly available risk assessment tools that identified risk factors among older adults. For the platform, functionality, and data security elements of the database, the decision to choose specific features was guided by the following principles: feasibility of implementing, cost of implementation, and possibility of scaling-up the technology.

Different questionnaires were evaluated by the expert panel to determine the contents of the database. The panel determined whether the tools met the following criteria: tested for reliability and validity in an older adult population, appropriate for the subsidized housing setting, could be delivered by volunteers, and feasible to be completed in a drop-in clinic setting. The different elements of the database (risk assessment tools and processes) were evaluated during weekly panel meetings for one month until all members came to a consensus on the features to be included in the initial database. The consensus building process involved evaluating the different tools against criteria, incorporating relevant ones into the database flow, then reviewing the database flow and making appropriate adjustments. This was repeated every week until consensus was established.

In Phase Two, the database was pilot tested with two mutually exclusive sets of older adults (five per set) and one set of paramedics. The pilot tests were done in different settings that were convenient to the participants. The settings included a research office building, the participants’ home, or a common space area in an apartment building. To evaluate the usability of the database, user experience was assessed. The variables studied included the following user-experience metrics: difficulty navigating the database, time to complete the database forms, and a prediction of whether paramedics and older adult volunteers will be able to complete the database forms. Participants were asked to enter their own health-related information or fictional information to complete the forms in the database. They subsequently completed a survey with 5-point Likert scale responses regarding their perceptions of the level of difficulty when using the database, the time needed to complete the forms, whether they thought paramedics or volunteer older adults would be able to complete this task, their familiarity with computers, their age, and any suggested changes to render the CP@clinic database more user-friendly. The questionnaire was developed to fit the purpose and elements of the database. This content was validated by the Expert panel. The results for each round of pilot testing were used to improve the database, which was then tested using the next group of participants.

Analysis

Data from the Phase Two survey was summarized using frequency distribution. The aim of the analysis was to evaluate the perceptions of older adults and paramedics regarding how user-friendly the database was.

Phase one: Database requirements and development

Through consultations with the expert panel and a review of the program documents, a set of database requirements were gathered and decisions were made regarding the database contents, platform, functionality, and data security. For each database element below, the identified requirements have been listed, followed by the solutions implemented.

Contents

The initial set of questionnaires presented to the expert panel included tools from the Canadian Community Health Survey (CCHS) [9], the Canadian Hypertension and Education Program (CHEP) [10], and the Canadian Diabetes Risk (CANRISK) questionnaire. Furthermore, since older adults living in subsidized housing had poor health status [1] and have frequently called 911 due to falls [11], health related quality of life questionnaires [12,13] (Euroqol 5 dimensions 3 levels) [14] and falls assessment (Timed-up-and-go or TUG test) [15] were added. The expert panel’s consensus of tools to be included in the database were as follows:

• Selected demographic variables (age, sex, ethnicity, marital status) and risk factor questions that were similar to what was collected in the CCHS (representative of the general population) so that comparisons can be made
• The CANRISK questionnaire as this was the only validated tool assessing diabetes risk in Canada
• Euroqol 5 dimensions 3 levels (EQ5D 3L) as this was the shortest tool for quality of life assessment and was used by health economists in Canada
• The TUG test since this was the standard tool recommended for falls assessment in older adults

Platform

The expert panel identified the need for the optimal database platform to be low-cost, secure, flexible (not dependent on internet access), capable of implementing complex algorithms and computations, and easy to use. The advantages and disadvantages for each of the database platforms considered are summarized in Table 1.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 1. Advantages and Disadvantages of different database platforms.

https://doi.org/10.1371/journal.pdig.0000689.t001

Accordingly, the computer-based database was selected since it was more flexible and less expensive, which was important for the sustainability of the CP@clinic program. This option did not require internet connections and gave researchers the ability to modify the database quickly.

Functionality

The database needed to be capable of multiple functions to support the efficient and effective implementation of CP@clinic. The database needed to capture initial and follow-up data entered by a user, perform calculations and summarize information to provide a risk assessment score, provide alerts to the user based on the results, compute participant age from the date of birth, the body mass index from the weight and height, and the CANRISK score, and to summarize all the participants’ risk factors, while guiding the user through the assessment process as they complete the forms and the risk assessment tools. Furthermore, once the participant’s risk was assessed, the database needed to provide resources available to manage risk factors that are tailored to the participant’s situation and generate automated reports for the researchers, participants, and the participants’ family physicians. To meet these database needs, the final format of the CP@clinic database initially used free software from the Centres for Disease Control, Epi Info 7 [16] but switched to Microsoft Access 2010 [17]. The database was structured like an electronic medical record, allowing cumulative information by participant.

Data security

The security of the data was a priority since the data contained personal health information. The database needed to be secure at the level of user access, the storage device (in case of theft or loss of the laptop), and data transmission (the data was transmitted to a secure project server). The CP@clinic data collected during sessions was locally stored in the CP@clinic laptop instead of directly in the CP@clinic central server to minimize the issue of poor internet connection in some locations. Moreover, the CP@clinic database is located in a hidden encrypted container/folder in the laptop using the Safehouse software [18]. The container and the database are password protected with a 20 character random alphanumeric password that was saved in a specialized Yubikey [19]. After each CP@clinic session, the paramedics return to the Hamilton Paramedic Service home base and are instructed to connect the laptop to their internal internet and synchronize the database to the CP@clinic central server using the FreeFileSync software [20]. The network folder on the central server saves the updated version of the CP@clinic database and maintains previous versions of the CP@clinic database. The CP@clinic central server is an SME Linux enterprise server, based on the CentOS operating system 7.5, that allows secure file sharing for authorized users [21]. The server is located on a secure university network, behind a university firewall, and is accessible via a virtual private network (VPN) connection for remote clients. The server is supported with nightly back-up onto a secondary onsite server with the same specifications and a third level secure off-site server backup. Access to the three servers is limited to approved individuals. In addition, users were unable to open unsafe web pages that could compromise the integrity of the operating software and all applications were locked one minute after the laptop was left unplugged and unused or when the laptop was in sleep mode. Lastly, anti-theft software, “Prey” [22], was installed to delete all files and take a snapshot of the person using the computer once the laptop was registered as stolen. All the software used in setting up the laptop system were freeware except for the anti-theft software, which had a minimal fee.

Phase two: Pilot testing the user-experience

Pilot test with older adults (set 1).

During the first pilot test, all five older adults found the database easy to complete and agreed that volunteer older adults would be able to complete the database. Although their experience with the database was generally positive, three of them preferred the option of dropdown boxes instead of textboxes when entering information (e.g., CANRISK scoring). They also had difficulty maneuvering the trackpad and preferred a mouse. Three of them indicated that they were “a little comfortable using computers”, while two said that they were “familiar” with using them. The time to complete data entry per participant ranged between three and thirteen minutes. Table 2 presents the survey results.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 2. Characteristics and perceptions of users regarding the CP@clinic database.

https://doi.org/10.1371/journal.pdig.0000689.t002

Research contributions

This study contributes to both academia and the healthcare sector by introducing a practical solution with versatile applications for similar community healthcare initiatives. The solution, characterized by its cost-effectiveness, portability, security, and user-friendliness, serves as a framework for enhancing healthcare accessibility and efficiency. This contributes to the advancement of patient-centered care practices in the health sector. Healthcare providers can deliver more efficient and effective care, ultimately improving health outcomes for this demographic. Moreover, the study’s findings suggest that the database solution created is scalable and replicable in comparable healthcare environments. The design is adaptable, which can accommodate various contexts and demographics, and the user-friendly nature of the database makes it accessible to a wide range of healthcare providers. The database’s capacity for scalability and adaptability implies that its influence may extend beyond its initial implementation. Academically, this study adds to the body of knowledge regarding effective strategies for managing and supporting older adults in community settings. It also adds to the body of knowledge of a pragmatic process of developing an electronic database with decision support for a community-based primary care program. This research not only addresses a gap between academia and practical healthcare implementation but also serves as a vehicle for improving patient outcomes and enhancing approaches to healthcare delivery in community settings.

Limitations

These results should be considered in the light of some limitations. The study was limited by the small sample size of participants used to test the database. Although we have variety in the range of prior technological knowledge and skills, this could be further strengthened by increasing the sample size. Moreover, our study focused on one geographical area, and thus it is difficult to generalize the results to all potential program areas. For instance, we did not have volunteers from rural/remote areas, who could have different experiences and perspectives regarding technology, database’s functionality and ease-of-use. Previous research has suggested variations in healthcare technology utilization and skills among healthcare providers and patients between rural and non-rural regions of Canada [26]. This implies a potential avenue for future research into the utilization of the database across different regions of Ontario.

This paper serves as a foundation for future exploration in database development for healthcare settings. There is a need for longitudinal studies to evaluate the impact and efficacy of the developed database solution over a longer period of time, alongside ongoing user feedback to refine it. Furthermore, investigations into the integration of new technologies like artificial intelligence or machine learning would be beneficial as it would provide insight into potential ways to enrich the database’s functionality and effectiveness. Finally, future research may focus on comparative analyses between the developed database solution and existing healthcare administration platforms or chronic disease management tools to reveal the database’s strengths and areas for improvement. By furthering this research, we can ensure that the database solution continues to evolve, maintaining its impact, accessibility, and utility. This, in turn, will improve healthcare delivery and lead to better outcomes for patients.