Correspondence | Published:

Genetics and Epigenetics

Information bias in measures of self-reported physical activity

International Journal of Obesity (2018) | Download Citation

In their recent genome-wide association study (GWAS) using UK Biobank, Klimentidis and colleagues [1] identified nine loci associated with self-reported physical activity (PA). We believe that rather than capturing genetic architecture of habitual PA, genetic variants identified in this GWAS, may at least in part reflect information bias in self-reported data, attributable to participants’ cognitive ability of reporting PA.

One of the strongest signals in this study was for APOE, with its risk allele being paradoxically associated with greater self-reported PA. Given the well-established association of APOE ε4 allele with Alzheimer’s disease [2], the direction of the association is unexpected. Although this could be attributed to selection bias as discussed by the authors (that is, APOE carriers in the study are enriched for healthy lifestyles), this association was not replicated using accelerometer-based PA. Although accelerometer data is only available for a subsample, the loss of power would be compensated by an even stronger selection bias operating in this subsample and objectively-assessed measures being less ‘noisy’. We believe that the more plausible explanation for this paradoxical association is that participants with lower cognitive function in the study tend to over-report their activity levels. This becomes more apparent in the genetic correlation analyses (which aggregate SNP effects across the genome), where increases in self-reported PA were found to be correlated with decreased intelligence, years of schooling, and lower childhood IQ, but increased risk for various mental-health related traits, including schizophrenia, PGC cross-disorder, Bipolar disorder, and Alzheimer’s disease. In contrast, none of these correlations surfaced in the analyses based on accelerometer-based PA, presumably as the objectively-assessed measures were not subject to the bias caused by participants’ cognitive ability of reporting PA.

To examine our hypothesis that participants with lower cognitive function in the study tend to over-report their activity levels, we tabulated average levels of self-reported PA against two distinct cognitive measures, participants’ age and education, and compared the patterns with those seen for accelerometer-based PA. As expected, participants with lower cognitive function, who were older, and who had lower levels of education reported higher levels of activity, with none or opposite patterns seen for the accelerometer-based data (Table 1). Further, this information bias in self-reported PA by indicators of cognitive function has also been reported in an independent sample [3], and in people with Schizophrenia within the UK Biobank [4]. In summary, associations with self-reported PA are likely biased by participants’ cognitive reporting ability, which should be considered in the context of these genetic findings, and which may limit their usefulness for subsequent genetic correlation or Mendelian Randomization analyses.

Table 1 Self-reported and objectively-measured physical activity by cognitive measures, age, and education


  1. 1.

    Klimentidis YC, Raichlen DA, Bea J, Garcia DO, Wineinger NE, Mandarino LJ et al. Genome-wide association study of habitual physical activity in over 377,000 UK Biobank participants identifies multiple variants including CADM2 and APOE. Int J Obes. 2018; 42: 1161–1176.

  2. 2.

    Liu C-C, Kanekiyo T, Xu H, Bu G. Apolipoprotein E and Alzheimer disease: risk, mechanisms and therapy. Nat Rev Neurol. 2013;9:106.

  3. 3.

    Herbolsheimer F, Riepe MW, Peter R. Cognitive function and the agreement between self-reported and accelerometer-accessed physical activity. BMC Geriatr. 2018;18:56.

  4. 4.

    Firth J, Stubbs B, Vancampfort D, Schuch FB, Rosenbaum S, Ward PB et al. The validity and value of self-reported physical activity and accelerometry in people with Schizophrenia: a population-scale study of the UK Biobank. Schizophrenia Bulletin. 2017; e-pub ahead of print 24 October 2017;

  5. 5.

    Bradbury KE, Guo W, Cairns BJ, Armstrong ME, Key TJ. Association between physical activity and body fat percentage, with adjustment for BMI: a large cross-sectional analysis of UK Biobank. BMJ Open. 2017;7:e011843.

  6. 6.

    Doherty A, Jackson D, Hammerla N, Plötz T, Olivier P, Granat MH, et al. Large scale population assessment of physical activity using wrist worn accelerometers: The UK Biobank Study. PLoS ONE. 2017;12:e0169649.

Download references

Author information


  1. Australian Center for Precision Health, University of South Australia Cancer Research Institute, Adelaide, Australia

    • Stephanie Folley
    • , Ang Zhou
    •  & Elina Hyppönen


  1. Search for Stephanie Folley in:

  2. Search for Ang Zhou in:

  3. Search for Elina Hyppönen in:

Conflict of interest

The authors declare that they have no conflict of interest.

Corresponding author

Correspondence to Elina Hyppönen.

About this article

Publication history