JMIR Publications

ABSTRACT

Background:

To be effective, digital health interventions (DHIs) have to be used. Accordingly, targeted application and/or design of DHIs requires knowledge of usage determinants. Thereby, two components of usage should be distinguished: uptake (using the DHI at least once) and frequency of use (extent of DHI use thereafter). Previous research has rarely examined these two components within the same study.

Objective:

The objective is to examine which determinants influence uptake and frequency of use, and whether these determinants differ between the two components.

Methods:

A secondary analysis of a randomised controlled trial is performed. This trial examined an intervention for improving disease-related self-management in persons with type 2 diabetes and/or coronary heart disease. The investigated DHI is a web portal provided in this intervention. The secondary analysis only refers to the intervention group. Age, gender, education, the Big Five, loneliness, depression, anxiety, health literacy, patient activation, and intentions regarding physical activity and healthy nutrition were examined as determinants. Each determinant was investigated separately using simple regression, and all determinants together using multiple regression. For uptake, logistic regression was applied and, for frequency, negative binomial regression with robust standard errors. Frequency was analysed for those who used the DHI at least once. Except for age, gender and education, analyses were conducted for variables standardised with 0 for the lowest and 1 for the highest possible measurement value. For simple regressions, inflation of alpha error due to multiple testing was controlled via the approach of Benjamini and Hochberg; for multiple regression via the significance of the complete multiple regression model.

Results:

Of the 462 intervention group members, 199 (43.1%) used the web portal at least once, with most log-ins recorded in the first six weeks. After controlling for inflation of alpha error, simple regressions for uptake yield significant effects for higher education (B=0.56; CI [0.18, 0.95]; p<.01); openness (B=1.08; CI [0.33, 1.83]; p<.01); intention regarding physical activity (B=2.28; CI [1.30, 3.26]; p<.001); and intention regarding healthy nutrition (B=2.30; CI [1.30, 3.31]; p<.001). The multiple regression model for uptake is highly significant (p<.001), with significant positive associations for intentions regarding physical activity (B=1.86; CI [0.74, 2.97]; p<.01) and healthy nutrition (B=2.22; CI [1.00, 3.44; p<.001), as well as a significant negative association for patient activation (B=-3.20; CI [ 4.95, 1.46]; p<.001). After controlling for inflation of alpha error, simple regressions for frequency yield no statistically significant effect, and the multiple regression model for frequency is not significant (p=.071).

Conclusions:

The examined variables only affected uptake but not frequency, suggesting different underlying determinants. High intentions regarding physical activity and healthy nutrition affected uptake positively, whereas high patient activation affected uptake negatively. Clinical Trial: The study was registered in the German Clinical Trials Registry (DRKS) in 2019 under the number 00020592.