Changes in physical activity during a ‘lockdown’

There is no shortage of investigations examining the impact of COVID-19 ‘lockdowns’ on physical activity. For example, a simple term search on PubMed using keywords of ‘physical activity’, ‘COVID-19’ and ‘lockdown’ returns 239 results since 2020 (search date: 5^th March 2021). Among these many results are preprints, original research, meta-analyses, systematic reviews and editorials. The consensus of this literature is that, indeed, physical activity declines during a COVID-19 ‘lockdown’. However, the notion that physical activity declines during a ‘lockdown’ presupposes that we know—with adequate confidence—the physical activity habits of the population before the ‘lockdown’ occurred.

Recently, Stockwell et al6 conducted a systemic review of all investigations that sought to determine changes in physical activity due to ‘lockdowns’ implemented throughout the COVID-19 pandemic. The overarching finding of this review was that physical activity declines and sedentary behaviour increases during a ‘lockdown’—these observations are seemingly consistent across multiple populations (eg, children, adults, clinical patient groups) and countries. It is noteworthy that among the 66 studies included in their systematic review, only four investigations objectively measured physical activity using wearable devices before and during a ‘lockdown’ period (eg, accelerometers, pedometers, activity trackers, etc). The remaining studies assessed physical activity levels using self-reported questionnaires. That is, by asking participants to recall their physical activity habits during a current ‘lockdown’ and at a time before the ‘lockdown’ began. Importantly, however, of the studies using questionnaires to assess physical activity, the majority were non-validated (n=39/59, 66%). Given that it is exceedingly difficult to predict when a COVID-19 ‘lockdown’ will occur, it is not surprising that most studies have relied on simple (although subjective) tools to assess physical activity in a retrospective, cross-sectional manner. It is however disconcerting that a large portion of the current evidence reporting a decline in physical activity during COVID-19 ‘lockdowns’ stems from investigations using non-validated tools. This problem is further compounded by the larger issue of asking participants to recall events that occurred in their lives over relatively long windows of time.

Recall bias is the systematic error that occurs when participants do not accurately remember or omit details of past life events and experiences. As such, recall bias is a serious threat to the internal validity of any self-reported questionnaire—a threat which increases as the period of recall is extended further back into the past. It must be remembered that even standardised questionnaires typically ask participants to recall their physical activity over the ‘previous 7 days’ or during a ‘typical week’—prompting for accurate information beyond this interval is tenuous. Further to this point, it is becoming clear that the on-going COVID-19 pandemic, and its many ‘lockdowns’, has adversely impacted on the mental health of the general public.7–9 It has been suggested that our ability to recall autobiographical events is influenced by our prevailing affective state (ie, mood-congruence recall bias).10 In this context, participants experiencing negative affect valence at the time of the survey may recall their physical activity habits prior to a COVID-19 ‘lockdown’ with greater pessimism than those experiencing a more positive affective state.

Is it just a matter of opinion?

The concerns outlined above raise the following questions: (i) What do bespoke (non-validated) surveys quantify if we are not certain of what they measure? (ii) How valid are standardised physical activity questionnaires when the period of recall is modified to refer to events occurring further in the past (ie, before a ‘lockdown’)?11 And finally: (iii) How might the mental health of a participant act to misconstrue or distort their recall of physical activity at any given point in time? At present, these questions do not have clear answers. Given that physical inactivity is already a pressing and serious concern for global health,5 12 it is important that we quantify the impact of the pandemic on physical activity habits as accurately as possible. Importantly, we do not suggest here that physical activity has not declined during the COVID-19 pandemic; rather, we argue that a conscious effort must be made by authors to acknowledge the specific limitations of their chosen self-reported measure of physical activity, such that we are appropriately cautioned to interpret the available evidence with consideration of the issues outlined above. Unless we can rule out the confounding effect of recall bias (and its moderators), we argue that the current literature may reflect a perceived rather than an actual change in physical activity during a COVID-19 ‘lockdown’. A participant’s recollection of their own physical activity habits before a ‘lockdown’ may be less a matter of fact, and more a matter of opinion.

Does it even really matter?

It cannot be denied that, as Scientists, it behoves us to find empirical evidence to support our theses. Within the context of this Viewpoint, such evidence would be offered by more investigations reporting on objective measures of physical activity during a COVID-19 ‘lockdown’. Until this evidence is brought to light, we must reconcile ourselves with the fact that most of our knowledge on the topic is derived from subjective measures of physical activity during a ‘lockdown’. But is that really a bad thing? If we reframe our thoughts, we see what is truly available to us: a large body of evidence showing that the general population perceives a decline in physical activity during a ‘lockdown’. And perhaps it is this belief that physical activity has declined which determines health outcomes at present, and into the future. For example, a self-perceived reduction in physical activity during the COVID-19 pandemic is seemingly associated with poorer mental health outcomes, as evidenced by higher stress and anxiety.13 According to this perspective, new questions arise: Does a perceived reduction in physical activity during a ‘lockdown’ encourage future sedentary behaviour once the pandemic recedes? Do we observe better mental health outcomes during, and post-‘lockdown’ if we can alter such perceptions of decreased physical activity? Answering these questions will help us to determine whether the public’s opinion of how the pandemic has affected our physical activity truly matters.