Adolescents’ food intake changes during the COVID-19 pandemic: The moderating role of pre-pandemic susceptibility, COVID-19 related stressors, and the social food context

ABSTRACT

Although insight in how adolescents’ food intake is affected by the COVID-19 pandemic is critical, knowledge is limited. Hence, this longitudinal study (N = 691, M_age = 14.30, SD_age = 0.62; 52.5% female) investigated changes in adolescents’ unhealthy (sugar-sweetened beverages, sweet snacks, savoury snacks) and healthy (fruit and vegetables) food intake (in total, obtained from home, and from outside the home) from pre-pandemic (Spring 2019) to the first lockdown (Spring 2020) and to six months later (Fall 2020). Moreover, several moderating factors were assessed. Results showed that the intake of unhealthy and healthy food in total and obtained from outside the home decreased during the lockdown. Six months later, unhealthy food intake returned to pre-pandemic levels, while healthy food intake remained lower. COVID-19 stressful life events and maternal food intake further qualified these longer-term changes in intake of sugar-sweetened beverages and fruit and vegetables, respectively. Future work is warranted to elucidate longer-term COVID-19 effects on adolescents’ food intake.

KEYWORDS:

• COVID-19 pandemic
• adolescence
• food intake
• longitudinal design

On 11 March 2020, the World Health Organization declared the outbreak of the COVID-19 virus a pandemic (WHO, 2020). Across the world, this pandemic and the measures aimed at containing the virus have profoundly affected individual’s daily lives. The measures may have particularly affected the lives of adolescents, as ‘a window of opportunity for attaining critical developmental tasks is temporarily closed’ (Keijsers et al., 2021, p. 117). More specifically, schools were closed, and adolescents were largely confined to their homes (RIVM, 2021). Additionally, their parents were at home more often, playing sports in a team of peers was not possible, school canteens, restaurants, and cafes were closed, and all parties and festivals were cancelled (RIVM, 2021), while adolescents were in the midst of developing important foundations for their future health and well-being. Studies suggest that the pandemic has affected several health behaviours (Chang et al., 2021; Viner et al., 2022). However, insight in how the pandemic has affected adolescents’ food intake specifically is still limited, while this knowledge is of utmost importance for public and individual health and well-being.

Adolescents’ food intake changes during COVID-19

It is critical for adolescents to adopt healthy food intake patterns to prevent the development of overweight in adolescence and beyond (Lien et al., 2001). Most optimally, these patterns include a low intake of nutrient-poor unhealthy food (e.g., sugar-sweetened beverages (SSBs), sweet, and savoury snacks; Larson & Story, 2013) and a high intake of healthy food (e.g., fruit and vegetables; Larson et al., 2007). There are concerns, however, that the COVID-19 pandemic and measures may have worsened adolescents’ food intake (Robinson et al., 2021). This can potentially be explained by mechanisms such as stress, boredom, or a lack of structure (Tomiyama, 2019) during the pandemic, triggering unhealthy, automatic food intake patterns instead of healthy habit development (Marteau et al., 2012). On the other hand, it is conceivable that the pandemic may have triggered healthier food intake patterns, for instance, through more parental monitoring and involvement (Cassinat et al., 2021) and limited exposure and access to (mostly unhealthy) food outlets outside the home environment (Timmermans et al., 2018), such as school canteens.

Since the outbreak of the pandemic, several studies have assessed the potential impact of the pandemic on adolescents’ food intake. More recently, these findings have been summarized in several review papers, indicating that a higher intake of both unhealthy food intake and healthy food has been reported during the first lockdown (Bennett et al., 2021; Mignogna et al., 2022; Stavridou et al., 2021). However, these studies were mostly limited by their cross-sectional designs, relying on retrospectively reported pre-pandemic food intake (Campbell & Wood, 2021). The few prospective studies to date show mixed findings, with some reporting increases (Pietrobelli et al., 2021, 2020), while others found decreases (Munasinghe et al., 2020; Perrar et al., 2022) in both healthy and unhealthy food intake. To elucidate whether, and in which direction, the COVID-19 pandemic has changed adolescents’ food intake in the Netherlands, we used a longitudinal design, comparing Dutch adolescents’ food intake before and during the pandemic. Additionally, we extended most previous work by not only assessing changes during the first lockdown (Spring 2020), but on the longer term as well (Fall 2020).

Moderating factors

In addition to better understanding potential changes in adolescents’ food intake, it is crucial to gain insight for whom the effects of the pandemic on food intake were most pronounced. The impact of the pandemic on adolescents’ unhealthy food intake may, for instance, be larger for adolescents who were already more susceptible to showing problematic weight-related behaviours before the pandemic. Consistent with this idea, previous cross-sectional studies among adults suggest that the effects of the COVID-19 pandemic on food intake are most pronounced among those with a higher body weight and more food addiction symptoms (Robinson et al., 2021; Schulte et al., 2022). We will uniquely examine the potential moderating effects of standardized Body Mass Index (zBMI) and food addiction prospectively in adolescents.

Moreover, as stress is generally related to weight gain and disordered eating patterns (Tomiyama, 2019), experiencing more stressful life events during the COVID-19 pandemic may (further) worsen adolescents’ food intake patterns. Moreover, not just the experience of these ‘objective’ stressors may be critical, subjective experiences (e.g., adolescents’ worries about getting infected by the COVID-19 virus) may be equally important, as more worrying has been previously associated with health behaviours (Kubzansky et al., 1997; Wright et al., 2021). As such, COVID-19 stressful life events and worries may moderate the effect of the pandemic on adolescents’ food intake.

Finally, the food intake of influential socializing agents of adolescents (e.g., parents and friends) may play a moderating role. Although prospective research assessing both parent and peer effects on food intake is limited in adolescence, it can be expected that friends gain importance, while parental influence decreases (Helsen et al., 2000). However, during the pandemic in general and during lockdowns in particular, adolescents spent more time at home (with parents) and less time outside the home (with friends; Rogers et al., 2021). Given these changes in the social context, the impact of the social food context on adolescents’ food intake could have changed as well.

The current study

Taken together, the first aim was to examine potential changes in Dutch adolescents’ food intake during the COVID-19 pandemic. Based on the existing cross-sectional studies, we expected that adolescents showed increased intake of both unhealthy (i.e., SSBs, sweet snacks, savoury snacks) and healthy (i.e., fruit and vegetables) food when comparing intake from before the pandemic to the first lockdown (Spring 2020, H1a) and to six months after the first lockdown (Fall 2020, H1b). In addition to total intake, we assessed intake obtained from home and from outside the home separately, to elucidate the generalizability of effects across food types and contexts.

Our second aim was to assess the moderating effects of pre-pandemic individual susceptibility, COVID-19 related stressors, and the social food context on these food intake changes. More specifically, we expected changes from pre-pandemic to the lockdown (H1a) and to six months after the first lockdown (H1b) to be more pronounced in adolescents with a higher pre-pandemic zBMI and food addiction (H2.1), in adolescents experiencing more COVID-19 stressful life events and more COVID-19 worries during the pandemic (H2.2), and in adolescents with mothers’ and best friends reporting a higher food intake (H2.3).

Method

Participants and procedure

The participants were part of the ‘G(F)OOD together!’ research project, a longitudinal observational cohort study on Dutch adolescents’ and their parents’ health behaviour (van den Broek et al., 2020). The participants in the current study were part of Wave 3 (Spring 2019, before the COVID-19 pandemic) and Wave 4 (Fall 2020, during the COVID-19 pandemic). In Wave 4, participants also retrospectively reported on the first Dutch lockdown (Spring 2020). These timepoints will from now on be referred to as pre-pandemic (Wave 3), lockdown (Wave 4 retrospectively), and follow-up (Wave 4). The procedures and data analyses for this paper were pre-registered (https://osf.io/n6m7z).

This study was approved by the Ethics Committee Social Sciences of Radboud University, Nijmegen, The Netherlands (ECSW20170805-516). An active parental consent procedure was used, and adolescents were asked to provide consent themselves as well. A total of 691 adolescents (pre-pandemic age: M = 14.30, SD = 0.62; 52.5% female) were included in the current study, of which 674 adolescents participated pre-pandemic, and 306 adolescents participated at follow-up (N = 289 adolescents participated in both timepoints).

Participants were informed that participation was voluntary, that answers would be processed anonymously, and that they could withdraw from the study at any moment. Pre-pandemic, adolescents completed a survey through Qualtrics Survey Software at school, and their height and weight were measured. At follow-up, due to the COVID-19 measures, adolescents completed an online survey without the researcher’s presence. Parents completed an online survey at home at both timepoints. At both timepoints, adolescents received a small gift for their participation, and thirty-four gift vouchers (values: 5 to 50 euros) and three weekend getaways (value: 250 euros) were raffled among the participating families.

Attrition analysis

A logistic regression analysis was performed to identify potential differences in demographics (i.e., adolescents’ gender, age, educational level, and zBMI) and main study variables (i.e., adolescents food intake) between those with missing (N = 385) and complete data (N = 306) at follow-up. The results (see Supplemental Table S1) revealed that attrition was only predicted by age (OR = 1.58, p < .001), indicating that older adolescents at pre-pandemic were more likely to have missing data at follow-up.

Measures

Adolescents’ food intake

To assess adolescents’ food intake, adolescents were asked to complete a food frequency questionnaire (FFQ) pre-pandemic, at lockdown, and at follow-up. Eight FFQ items were combined to form four food constructs. More specifically, scores for soft drinks (FFQ item 1) were used to obtain the measure for SSBs. Scores for cake, candy bars, and chocolate (FFQ items 2 to 4) were summed to assess sweet snacks. Scores for warm, fried snacks (FFQ item 5) were used to assess savoury snacks. Scores for fruit, salad and raw vegetables, and heated vegetables (FFQ items 6 to 8) were summed to assess fruit and vegetables. Our rationale for and formulation of these items are presented in our pre-registration (https://osf.io/n6m7z) and previous work (Van den Broek et al., 2020).

For all FFQ items, adolescents were asked to indicate on how many days per week (score 0–7) they obtained this item in four different contexts: (1) taken or received from home, to eat or to drink at home or to take away; (2) bought at school, such as from the cafeteria or the vending machine; (3) bought somewhere else, such as in the supermarket, snack bar, or sports club; and (4) received somewhere else, such as at their neighbours’, grandparents’, or friends’ place. Responses to all four contexts were summed to obtain adolescents’ total intake scores. As done previously (van den Broek et al., 2020), a measure representing intake of food obtained from outside the home for each FFQ item was constructed by summing the responses on contexts 2 to 4.

zBMI

First, adolescents’ BMI was computed by dividing their weight in kilograms by their squared height in metres, both measured pre-pandemic. Subsequently, adolescents’ zBMI was computed by considering the age- and gender-specific growth curves for BMI, based on a Dutch representative sample of 0-to-21-year-olds (Schönbeck et al., 2011).

Food addiction

The modified Yale Food Addiction Scale (Flint et al., 2014) for children (Gearhardt et al., 2013) administered at Wave 2 (Spring 2018) was used. We computed a dimensional score (Schiestl & Gearhardt, 2018) by summing the scores on the 7 Likert-scale items (e.g., ‘When I start eating, I find it hard to stop.’), to which adolescents could respond on a 5-point scale ranging from ‘Never’ (score 0) to ‘Always’ (score 4). The scale showed good internal consistency (α = .71).

COVID-19 stressful life events

We developed a COVID-19 stressful life events scale for our project, by adjusting an existing checklist on general stressful life events (Garnefski et al., 2001). While we also asked to report on pre-pandemic life events, we only included the six stressful life events reported during the COVID-19 pandemic (March 2020 until follow-up (i.e., Fall 2020)) in the current study (e.g., ‘serious conflicts/quarrels in the family’; see Supplemental Table S2). The event was scored ‘1’ if the event occurred, and ‘0’ if the event did not occur. Subsequently, a sum score was created.

COVID-19 worries

To assess COVID-19 worries, we used two items at follow-up developed by the COVID-19 International Student Well-being Study (van de Velde et al., 2021) to ask separately about worries of 1) him/herself and 2) a family member getting (re-)infected by the COVID-19 virus. Adolescents could respond on a visual analogue scale ranging from ‘very worried’ (score 0) to ‘not worried at all’ (score 100). The two items were averaged to obtain one COVID-19 worries score (α = .76).

Mother’s food intake

A total of 344 biological mothers of the participating adolescents participated pre-pandemic. Pre-pandemic, for each of the eight FFQ items also administered to adolescents, mothers were asked to indicate on how many days per week (score 0–7) they consumed the different food items: (1) in the presence of their child and (2) in the absence of their child. The eight FFQ items were combined into the four food constructs, and the intake in presence and absence of the child were summed to create a total score (van den Broek et al., 2020).

Best friends’ food intake

To assess best friends’ food intake, adolescents were asked pre-pandemic to nominate their best friend from their grade at school. They were allowed to nominate same- and other-gender classmates, but were not allowed to choose themselves. Once a best friend was identified, this individual’s total food intake score was matched to the adolescents’ scores, to use as their best friends’ food intake measure (van den Broek et al., 2020). Of the 674 participating adolescents pre-pandemic, 612 nominated an identifiable best friend from their grade. Of the identifiable nominations, 342 scores could be matched.

Covariates

Adolescents’ pre-pandemic age in years was derived from their date of birth and date of participation. Educational level was scored as 1 = lower general secondary education, 2 = higher general secondary education, and 3 = pre-university education. Adolescents’ gender was coded as 0 = male and 1 = female.

Strategy of analysis

To test our hypotheses, we performed linear mixed-effects models (lme4 package; Bates et al., 2021) in R (R Core Team, 2022), to account for the nesting of repeated measures within participants and of participants in classrooms. All models were run separately for the four types of adolescents’ food intake (SSBs, sweet snacks, savoury snacks, and fruit and vegetables) in the three contexts (total, obtained from home, obtained from outside the home). All continuous predictors were centred, and gender was sum-to-zero coded (males = 1 and females = −1). Two ‘simple’ time contrasts were created to assess changes from pre-pandemic to (a) lockdown and to (b) follow-up. The exact model specifications can be found in our R script (https://osf.io/wqg5h/).

Statistical significance of model estimates was determined by obtaining both p-values (Type 3 conditional F-tests with Kenward-Roger approximation for degrees of freedom, afex package; Singmann et al., 2021) and confidence intervals (99.6%, bootstrapped). To interpret significant interaction effects, we performed post-hoc comparisons (emmeans package; Lenth, 2022) at low (−1SD), average (M), and high (+1SD) levels of the moderators. Model residuals and influential cases were inspected, after which models with significant effects of interest were performed without outliers (scaled absolute residuals > 3) and/or influential cases (Cook’s distance > 4/N, influence.ME package; Nieuwenhuis et al., 2012) as sensitivity analyses.

Deviations from pre-registration

We deviated in two ways from our pre-registered procedures. First, as we performed overlapping tests, we decided to correct for multiple testing. The pre-registered p-value of .05 was divided by 12 (3 overlapping unhealthy food constructs * 2 overlapping moderators * 2 overlapping time tests), resulting in using a corrected p-value threshold of .004 (and of a 99.6% confidence interval, accordingly). Second, we used mothers’ and best friends’ food intake measured before, not during, the pandemic as moderators, as it makes more statistical and theoretical sense that the moderator proceeds change on the dependent variable.

Results

Descriptive statistics

In Supplemental Figures S1 to S4, descriptive mean-based trajectories of food intake at pre-pandemic, lockdown, and at follow-up are presented. Furthermore, descriptive statistics of the moderator variables are shown in Supplemental Table S3.

Model H1: Adolescents’ food intake changes during COVID-19

Our results (Table 1, H1a) showed that across almost all types of food intake, total intake and intake obtained from outside the home decreased significantly during lockdown. An exception was sweet snacks, of which total intake did not significantly decrease. Moreover, intake obtained from home increased for sweet snacks and decreased for fruit and vegetables, although these findings were not confirmed by sensitivity analyses (Table 1, estimates between brackets). When comparing pre-pandemic to follow-up, results (Table 1, H1b) indicate that intake of SSBs, sweet and savoury snacks returned to pre-pandemic levels in all contexts, while fruit and vegetables intake remained lower across all contexts.

Table 1. Unstandardized Estimates of Mixed-Effects Models for the Intake of SSBs, Sweet Snacks, Savoury Snacks, and Fruit and Vegetables (N = 691).

SSBs

Sweet snacks

Savoury snacks

Fruit and vegetables

Total

Home

Outside Home

Total

Home

Outside Home

Total

Home

Outside Home

Total

Home

Outside Home

Model H1

Pre-pandemic vs. lockdown (H1a)

−0.59 [−0.53]

0.17

−0.78 [−0.81]

−0.83

0.66 [0.24]

−1.51 [−1.42]

−0.29 [−0.34]

0.06

−0.36 [−0.34]

−1.66 [−1.52]

−0.59 [−0.41]

−1.11 [−0.97]

Pre-pandemic vs. follow-up (H1b)

−0.08

−0.03

−0.09

−0.26

0.19

−0.46

−0.09

−0.09

−0.01

−1.50 [−1.30]

−0.62 [−0.48]

−0.92 [−0.71]

Model H2.1

H1a X zBMI

−0.27

−0.32

0.06

0.26

−0.02

0.29

0.02

−0.03

0.06

−0.03

−0.19

0.15

H1b X zBMI

<0.01

−0.14

0.15

0.40

0.18

0.22

0.01

<0.01

0.02

−0.30

−0.22

−0.08

H1a X food addiction

0.02

0.08

−0.06

−0.23

−0.05

−0.18

−0.03

<0.01

−0.04

0.03

0.13

−0.11

H1b X food addiction

−0.04

0.04

−0.08

−0.18

−0.04

−0.14

−0.04

<0.01

−0.04

0.05

0.12

−0.08

Model H2.2

H1a X life events

0.13

0.09

0.02

−0.02

0.03

−0.05

0.19

0.11

0.09

−0.26

−0.30

0.05

H1b X life events

0.66 [0.42]

0.27

0.39 [0.33]

0.44

0.05

0.39

0.15

0.06

0.10

−0.10

−0.17

0.08

H1a X worries

0.02

0.01

0.02

0.09

−0.08

0.17

0.02

0.01

0.01

−0.06

−0.11

0.05

H1b X worries

−0.07

−0.08

0.01

0.05

−0.06

0.11

0.06

0.02

0.03

−0.04

−0.06

0.02

Model H2.3

H1a X mothers’ intake

−0.09

−0.07

−0.02

−0.19

−0.09

−0.10

−0.18

−0.01

−0.15

0.13

0.10

0.03

H1b X mothers’ intake

0.06

−0.01

0.07

−0.09

−0.04

−0.06

−0.22

−0.19

<0.01

0.13

0.15 [0.12]

−0.01

H1a X friends’ intake

0.01

−0.02

0.03

0.01

0.02

−0.02

0.11

0.08

0.02

−0.08

−0.06

−0.04

H1b X friends’ intake

0.16

0.01

0.16

0.02

−0.01

0.03

0.13

0.03

0.08

−0.14

−0.11

−0.06

Statistically significant effects (corrected p-value < .004) are denoted in bold. Exact significance statistics are presented in Supplemental Tables S4 to S7. Sensitivity analyses for statistically significant effects are denoted between brackets (in bold if p-value < .004). SSBs = Sugar-sweetened beverages. zBMI = standardized Body Mass Index. All models are controlled for adolescents’ gender, educational level, and age.

Model H2.1 to H2.3: Moderating factors

There were no statistically significant interaction effects for zBMI and food addiction symptoms (Table 1, Model H2.1). There were also no other factors significantly moderating effects on sweet snacks and savoury snacks intake (Table 1, Model H2.2 and Model H2.3).

However, COVID-19 stressful life events significantly moderated adolescents’ intake of SSBs in total and obtained from outside the home from pre-pandemic to follow-up (Table 1, Model H2.1). Post-hoc comparisons indicated that while total SSBs intake decreased for adolescents who experienced few life events (b = −0.50, SE = 0.25, p = .044), intake increased for adolescents who experienced many life events (b = 0.79, SE = 0.18, p = .002), and did not change for those experiencing the average amount of life events (b = 0.15, SE = 0.26, p = .409), see Supplemental Figure S5. The interaction pattern was similar for the intake of SSBs obtained from outside the home (low: b = −0.29, SE = 0.16, p = .082; average: b = 0.10, SE = 0.12, p = .400; high: b = 0.48, SE = 0.17, p = .004), see Supplemental Figure S6. Sensitivity analyses only replicated the moderation on the intake of SSBs obtained from outside the home (Table 1, estimates between brackets).

Additionally, it was found that mother’s fruit and vegetables intake moderated adolescents’ intake of fruit and vegetables obtained from home from pre-pandemic to follow-up (Table 1, Model H2.3). Post-hoc comparisons indicated that while fruit and vegetables intake obtained from home significantly decreased for adolescents whose mothers consumed low (b = −2.52, SE = 0.59, p < .001) and average (b = −1.31, SE = 0.40, p = .001) amounts of fruit and vegetables, it did not change for adolescents whose mothers consumed high amounts of fruit and vegetables (b = −0.10, SE = 0.56, p = .861), see Supplemental Figure S7. This finding was not confirmed by the sensitivity analyses (Table 1, estimates between brackets).

Discussion

This study aimed to investigate changes in adolescents’ food intake during the COVID-19 pandemic, and potential moderating effects of pre-pandemic individual susceptibility, COVID-19 related stressors, and the social food context. For both healthy and unhealthy food intake, total intake decreased during the first lockdown, mainly due to declines in the intake of food obtained from outside the home. Although non-significant in the sensitivity analyses, intake obtained from home increased for sweet snacks and decreased for fruit and vegetables. On the longer term, the intake of unhealthy food returned to pre-pandemic levels, while fruit and vegetables remained lower across all contexts. Longer-term changes in SSBs intake obtained from outside the home and fruit and vegetables intake obtained from home were further qualified by the amount of experienced COVID-19 stressful life events and their mother’s food intake, respectively.

The decrease of both healthy and unhealthy food intake observed for total intake and intake obtained from outside the home during the first lockdown is in line with other prospective studies showing decreases in youth’s total energy (Perrar et al., 2022) and adolescents’ fast-food consumption (Munasinghe et al., 2020). These results are, however, not in line with cross-sectional studies, which showed higher intake of both unhealthy food and fruit and vegetables during the first lockdown (Bennett et al., 2021; Mignogna et al., 2022; Stavridou et al., 2021). These discrepancies can be the result of cross-sectional designs relying on retrospectively reported food intake before the pandemic. Our and the previous prospective findings seem to suggest that retrospectively reported pre-pandemic food intake measurements have been limited by recall biases (Cross et al., 2021; Hipp et al., 2020).

Based on these prospective findings of the first lockdown, some researchers have stated that these results are encouraging (Perrar et al., 2022), as the pandemic does not seem to promote increases in unhealthy food intake among adolescents as was initially assumed (Robinson et al., 2021). Our findings call for a more nuanced view, by showing that these short-term lockdown effects on unhealthy food intake were reversed when measures were (temporarily) relaxed. More specifically, we found that six months after the first lockdown, when measures were more relaxed, the intake of unhealthy food returned to pre-pandemic levels in all contexts. These findings seem to indicate that while changing the environment temporarily can reduce adolescents’ unhealthy food intake, these effects disappear when adolescents are confronted again with the (largely unhealthy; Timmermans et al., 2018) food environment outside their home (de Vet et al., 2013). It can be speculated that for healthy food like fruit and vegetables, the decreases induced by the lockdown ‘home context’ have become automatized, and are not reversed again by the environment outside the home (i.e., taking fruit to school). It is likely that adolescents do not associate the out-of-home environment (e.g., at school, with peers) with the intake of fruit and vegetables, but rather to eat unhealthy food (Stok et al., 2016).

Notably, while food intake changes from pre-pandemic to lockdown seem universal across adolescents, food intake did in some cases depend on other factors when considering changes to six months after the lockdown. First, while SSBs intake was below or back at pre-pandemic levels for adolescents who experienced relatively few to average life events, it was significantly higher than pre-pandemic levels for adolescents experiencing relatively many life events. This suggests that not only stress in general (Tomiyama, 2019), but also COVID-19 related stress can lead to increases in intake of unhealthy food, specifically SSBs. The fact that this effect is limited to the out-of-home context during follow-up may reveal that adolescents seem to deal with these stressors outside the home when measures were (temporarily) relaxed and adolescents were allowed to do things outside the home again, for instance, with friends. This potential explanation is further supported by the fact that most stressful life events assessed in this study were related to the family/home environment (see Supplemental Table S2). Second, while fruit and vegetables intake remained at decreased levels for adolescents whose mothers consumed relatively low amounts of fruit and vegetables, it was similar again to pre-pandemic levels for adolescents whose mothers consumed relatively high amounts of fruit and vegetables. As such, we suggest that parental modelling and home environmental structuring might still play a role beyond childhood, among (relatively older) adolescents. We should be careful, however, with the interpretation of this specific finding as it was not replicated in the sensitivity analyses.

This study had several strengths. First, we used a prospective research design with data available from both before and during the pandemic. As such, we could provide a less biased estimate of the effects of the pandemic on adolescents’ food intake. Uniquely, we were one of the first (Pietrobelli et al., 2021) to not only include data during the first lockdown, but on the longer term as well when measures were largely relaxed, providing an indication of adolescents’ resilience. Second, we used a range of foods (both healthy and unhealthy) across different contexts (both obtained from home and from outside the home). This elucidates the generalizability (versus domain-specificity) of the impact on adolescents’ food intake across food types and contexts, which is informative to pinpoint which specific behaviours and contexts interventions should focus on. Third, when assessing mothers’ and best friends’ food intake as moderators, we obtained their own self-reports, to not rely on adolescents’ perceptions of the food intake of these potential socializing agents (Kandel, 1980).

Our study also had some limitations. First, we used self-reports, which can be subject to limitations, including social desirability bias (Barros et al., 2003). Second, although we used a longitudinal design, we cannot infer causality. Notably, we cannot disentangle normative age-effects from COVID-19-specific effects, although our findings are largely in line with our recent study in which normative age-effects were eliminated using a cohort-comparison (Van den Broek et al., 2022). Third, it is a limitation that the first lockdown (Spring 2020) was assessed by retrospective reports during the Fall 2020 measurement. However, this retrospective report was done during the COVID-19 pandemic itself, making reports less subject to context-dependent distorted memory processes (Hupbach et al., 2008; Smith & Vela, 2001). Fourth, we used number of days per week as our food intake measure, while others also considered amounts consumed per day. It is possible that while the number of days at which a product consumed decreases, amounts consumed per day increased. Finally, although there are commonalities in the COVID-19 measures taken across countries (e.g., school closures), the specific restrictions and the timing of the measures differed between countries. As such, our findings may not fully generalize to adolescents in countries outside of the Netherlands.

Adolescence is a time of transition in which greater independence from parents is achieved, which includes increases in autonomy in obtaining, preparing, and consuming of food (Neufeld et al., 2022). During the COVID-19 pandemic, adolescents’ opportunities for achieving independence, including their food intake, may have been limited. This study concludes that during this unique period of time, total healthy and unhealthy food intake decreased significantly during the first lockdown, while the intake of unhealthy food returned to pre-pandemic levels in all contexts when measures were (temporarily) relaxed. Fruit and vegetables intake remained lower across all contexts. This may emphasize the importance of the environmental (social) context outside the home on adolescents’ increasing unhealthy food intake patterns. Moreover, as fruit and vegetables intake obtained from home tended not to decrease in adolescents whose mothers consumed high amounts of fruit and vegetables, and since SSBs intake decreased in adolescents who experienced few stressful life events, interventions could for instance, focus on promoting healthy food intake in parents and help adolescents cope with stress in their home environment. Future work is warranted to further elucidate the longer-term effects of the COVID-19 pandemic on adolescents’ food intake, particularly given that lockdowns and strict measures were reinstated to varying degrees for at least until one year after our follow-up measurement, likely further and potentially more chronically affecting adolescents’ food intake patterns.

Acknowledgments

We would like to thank all the participating schools, adolescents, and parents for their contribution to this research project. Moreover, we would like to thank all the student assistants for their help during the data collection of this project.

Disclosure statement

No potential conflict of interest was reported by the author(s).

Data availability statement

The data used for the current study are available upon request from the corresponding author.

Supplementary material

Supplemental data for this article can be accessed online at https://doi.org/10.1080/17405629.2022.2115999

Additional information

Funding

This research was supported by ZonMw under Grant Number 10430032010009