An evaluation of patients' experienced usability of a diabetes mHealth system using a multi-method approach

mHealth systems are becoming more common to aid patients in their diabetes self-management, but recent studies indicate a need for thorough evaluation of patients' experienced usability. Current evaluations lack a multi-method design for data collection and structured methods for data analyses. The purpose of this study was to provide a feasibility test of a multi-method approach for both data collection and data analyses for patients' experienced usability of a mHealth system for diabetes type 2 self-management.

A random sample of 10 users was selected from a larger clinical trial. Data collection methods included user testing with eight representative tasks and Think Aloud protocol, a semi-structured interview and a questionnaire on patients' experiences using the system. The Framework Analysis (FA) method and Usability Problem Taxonomy (UPT) were used to structure, code and analyze the results. A usability severity rating was assigned after classification.

The combined methods resulted in a total of 117 problems condensed into 19 usability issues with an average severity rating of 2.47 or serious. The usability test detected 50% of the initial usability problems, followed by the post-interview at 29%. The usability test found 18 of 19 consolidated usability problems while the questionnaire uncovered one unique issue. Patients experienced most usability problems (8) in the Glucose Readings View when performing complex tasks such as adding, deleting, and exporting glucose measurements. The severity ratings were the highest for the Glucose Diary View, Glucose Readings View, and Blood Pressure View with an average severity rating of 3 (serious). Most of the issues were classified under the artifact component of the UPT and primary categories of Visualness (7) and Manipulation (6). In the UPT task component, most issues were in the primary category Task-mapping (12).

Multiple data collection methods yielded a more comprehensive set of usability issues. Usability testing uncovered the largest volume of usability issues, followed by interviewing and then the questionnaire. The interview did not surface any unique consolidated usability issues while the questionnaire surfaced one. The FA and UPT were valuable in structuring and classifying problems. The resulting descriptions serve as a communication tool in problem solving and programming. We recommend the usage of multiple methods in data collection and employing the FA and UPT in data analyses for future usability testing.