The utility of physical activity questionnaires among African origin populations; lost in translation?

Background: Physical activity (PA) is an important preventive factor of non-communicable diseases (NCDs), particularly cardiovascular disease, yet progress towards reducing physical inactivity in populations is slow. Population-levels of PA are most often estimated using self-report questionnaires in population surveys, such as the Global PA Questionnaire (GPAQ), which may not accurately reflect objectively measured PA, such as accelerometers. The aim of the current study was to compare self-report vs objectively measured PA across 5 African-origin populations.

Methods: Approximately 2,000 African-origin men and women (35-55 yrs) were enrolled in a prospective cohort study to explore the relationship between lifestyle and cardiometabolic health. Participants were from Ghana, South Africa (SA), Jamaica, Seychelles, and the United States (US). Data collection included objective PA (accelerometer) and self-reported PA (GPAQ). WHO defines "sufficiently active" as performing >150 minutes of moderate intensity PA per week, or >600 MET*min/week (metabolic equivalent of task in minutes/week). Logistic regression was used to determine the proportion of participants that were physically active by accelerometry, among those who were found as sufficiently active based on GPAQ.

Results: 1,161 participants had complete self-reported PA and accelerometery data. Overall, 23.5% were classified as sufficiently active by both PA measures, while 38.2% of those classified as sufficiently active based on self-reported PA did not meet the criteria using objective monitoring (sensitivity of GPAQ: 74%, specificity: 44%). Among participants who were classified as sufficiently active according to the questionnaire (n=717), participants from Ghana (OR=3.2, p<0.01), SA (OR=5.3, p<0.01), Jamaica (OR=2.3, p=0.01), and Seychelles (OR=1.9, p=0.03) were more likely to be similarly classified as sufficiently active based on accelerometry, compared to those from the US, and men (OR=2.9, p<0.01) more likely than women. Finally, obese participants had 0.4 times lower adjusted odds to be similarly classified as sufficiently active using objective measures compared to non-obese participants (p<0.01).

Conclusion: Our findings further underscore difficulties in interpreting self-reported PA, which may furthermore vary across different settings and socio-economic settings. Given the importance of PA interventions for reducing the NCD burden, future research should explore tailored approaches to better understand self-report PA accuracy and to better assess the dose-response between PA and cardiometabolic risk.