Development of a New Cell Phone-Based Physical Activity Recall Using Daily Activity Pictures

Purpose Self-reported physical activity (PA) has several challenges, including recall errors, social desirability, incomplete answers and misunderstanding of questionnaires. All of these affect PA estimates. To reduce recall errors, daily PA diaries have been used as a criterion for other types of PA questionnaires. The purposes of this study were to 1) develop a cell phone-based PA recall using periodic photographs and 2) compare reported minutes of moderate PA (MPA) per day using the International Physical Activity Questionnaire (IPAQ), cell phone-based PA recall using their activity pictures (named, PA recall), and ActiGraph GT3X accelerometer.

Methods Each participant wore a GT3X monitor for 3 days. During these 3 days, participants used their cell phone to take 3 pictures (in the morning, afternoon, and evening) per day of their daily activities for a total of 9 pictures. Participants were given the IPAQ within a week, and a PA recall (their minutes and intensity level of PA) was then conducted using their 9 pictures. After a PA recall, the IPAQ was given again. Summary statistics found the skewness of data, prompting the use of the Kruskal- Wallis and Wilcoxon Rank-Sum tests. Three comparisons were made: 1) between PA recall, ActiGraph, and IPAQ before PA recall; 2) between PA recall, ActiGraph, and IPAQ before and after PA recall; and 3) between IPAQ before and after PA recall. Significance level was set P less than 0.05. All analyses were performed using Stata version 11.2.

Results Thirteen participants, 10 females and 3 males, mean age 23.3 ± 6.14 yr took daily activity pictures using cell phones. The mean scores of MPA by PA recall, IPAQ, and ActiGraph were 24, 26, and 31, minutes per day, respectively. The Kruskal- Wallis test yielded that there is no difference in minutes of MPA per day between the Actigraph, IPAQ, and PA recall, (p= 0.086), and no difference between the Actigraph, IPAQ (after PA recall), and PA recall (p=0.056). The Wilcoxon rank-sum test yielded there was no difference between before and after PA recall in IPAQ MPA minutes. (p= 0.064)

Conclusions The cell phone-based PA recall resulted in similar estimates of MPA by reported ActiGraph and IPAQ. The first generation of cell phone-based PA recall using activity pictures is well accepted and may help to predict better estimates of PA by reducing recall errors.