An observational cross-sectional study design was implemented. An open, on-line, self-administered survey was used to collect data on socio-demographics and mental health. Data on mobility was collected from Google Community Mobility Reports.14 Data collection took place between May and June 2020, at that time confinement in Ecuador was mandatory.15

Our targeted population were young adults that is, individuals between 18 and 35 years old, men and women. Invitations to participate in the study were published in platforms and social networks (i.e. Facebook, Instagram, Twitter and WhatsApp) of several Ecuadorian universities. In this way, the study population came from a non-probabilistic sampling and was mostly composed by university students and teachers. From the 9582 responses received, we exclude people younger than 18 or older than 35 years old, whose age was not recorded or whom selected “other” as their gender (n=1156). The final sample included 8426 young adults.

The survey was reviewed by four experts and first applied to 30 adults as a pilot study. After corrections were implemented, it was then distributed through social networks using Google Forms. It was mandatory that participants filled out some of the questions and no incentive was offered for answering the survey. Participants could review their responses and change their answers through a back button. Duplicate entries were avoided by preventing users access twice.

We implemented four linear multiple regression models, one for each of the mental health aspects that were evaluated. In this way, scores for emotional eating, sleep quality, depression and sense of coherence were included in the regression models as dependent variables. Our main predictors were percentage of mobility change, gender and age. From the six aspects of mobility initially considered, we included two in the models in order to avoid collinearity, since most of them were highly correlated (Supplementary Fig. I). These included mobility to workplaces and to groceries/pharmacies, which were of interest for our purpose.

All analyses were conducted using R version 3.6.3 (2020-02-29),20 RStudio version 1.3.1093,21 and the package rms to implement the regression models.22 Adjusted predicted means with 95% confidence intervals (95%CI) were calculated for each of the scores. Predicted means were adjusted by living with a partner, taking sleeping pills, area of residence (urban vs. rural), occupation (student, working or unemployed), and auto-report of times per week of leaving home. Age was included in the model as a linear predictor with restrictive cubic splines using the function “rcs”. Bootstrap estimates of the covariance matrix of coefficients were calculated using the function “bootcov” to obtain unconditional estimates. We applied 500 repetitions. Bootstrap estimates were corrected for intra-cluster correlation. A cluster was considered to be the province the respondents reported to be living in.

The present study was carried out following the Declaration of Helsinki to work with human beings and in accordance with the Singapore Declaration on Integrity in Research. It was approved by the Scientific Committee of the School of Medicine and Instituto de Investigación (IDI) of the Escuela Superior Politécnica de Chimborazo. Participants who voluntarily accepted to participated, filled out an informed consent form, which was included at the beginning of the survey. We did not collect any personal information that allowed identification.

Both, mobility change to workplaces (df=1, F=16.40, p<0.001) and groceries/pharmacies (df=1, F=4.64, p=0.031) were significantly associated to emotional eating behavior. In this way, a higher restriction of mobility for both places was associated with less healthy eating behavior. Emotional eating behavior was also significantly associated to gender (df=1, F=186.67, p <0.001) and age (df=4, F=25.24, p <0.001). For instance, women as well as youngsters showed significantly less healthy eating behavior than men and older respondents (Fig. 1 and Supplementary Fig. II). Model information, adjusted predicted emotional eating behavior scores and their 95% confidence intervals for predictors are showed in Supplementary Table I.

Figure 1.

Eating behavior by age, gender and mobility. Predicted means are showed for mobility to workplaces. Red line indicates higher restriction (3rd quantile) and the blue lower restriction (1st quantile). Confidence intervals were omitted to facilitate visualization, but can be consulted in Supplementary Table I.

(0.11MB).

Both, mobility change to workplaces (df=1, F=13.41, p <0.001) and groceries/pharmacies (df=1, F=7.93, p=0.005) were significantly associated to depression. In this way, a higher restriction of mobility for both places was associated with a higher score for depressive symptoms. Depressive symptoms were also significantly associated to gender (df=1, F=69.10, p <0.001) and age (df=4, F=10.78, p <0.001). For instance, women as well as youngsters more often showed depressive symptoms than men and older respondents (Fig. 2 and Supplementary Fig. III). Model information, adjusted predicted PHQ-9 scores and their 95% confidence intervals for predictors are showed in Supplementary Table I.

Figure 2.

Depression by age, gender and mobility. Predicted means are showed for mobility to workplaces. Red line indicates higher restriction (3rd quantile) and the blue lower restriction (1st quantile). Confidence intervals were omitted to facilitate visualization, but can be consulted in Supplementary Table I.

(0.1MB).

Both, mobility change to workplaces (df=1, F=13.41, p <0.001) and groceries/pharmacies (df=1, F=7.93, p=0.005) were significantly associated to sleep quality. In this way, a higher restriction of mobility for both places was associated with a worse sleep quality. Sleep quality was also significantly associated to gender (df=1, F=7.48, p=0.006), but not to age (df=4, F=1.98, p=0.094). In comparison to men, women showed higher PSQI than men, suggesting worse sleep quality (Fig. 3 and Supplementary Fig. IV). Model information, adjusted predicted PSQI scores and their 95% confidence intervals for predictors are showed in Supplementary Table I.

Figure 3.

Sleep quality by age, gender and mobility change. Predicted means are showed for mobility to workplaces. Red line indicates higher restriction (3rd quantile) and the blue lower restriction (1st quantile). Confidence intervals were omitted to facilitate visualization, but can be consulted in Supplementary Table I.

(0.11MB).

Both, mobility change to workplaces (df=1, F=4.31, p=0.038) and groceries/pharmacies (df=1, F=4.03, p=0.045) were significantly associated to sense of coherence. In this way, a higher restriction of mobility for both places was associated with a lower score for sense of coherence. Sense of coherence was also significantly associated to age (df=4, F=67.79, p <0.001) but not to gender (df=1, F=1.57, p=0.210). Youngsters showed significantly lower sense of coherence than older respondents (Fig. 4 and Supplementary Fig. V). Model information, adjusted predicted SOC-15 scores and their 95% confidence intervals for predictors are showed in Supplementary Table I.

Figure 4.

Sense of coherence by age, gender and mobility change. Predicted means are showed for mobility to workplaces. Red line indicates higher restriction (3rd quantile) and the blue lower restriction (1st quantile). Confidence intervals were omitted to facilitate visualization, but can be consulted in Supplementary Table I.

(0.13MB).

The fact that lock-down measures negatively influence mental health is supported by previous evidence.23,24 However, the present is one of the few that took into account compliance to confinement, although indirectly measured. Among young adults, factors associated with higher mental distress due the confinement might be related not only to their inner psychological characteristics but also to adverse socioeconomic circumstances derived from the pandemic.24–26 In this line, previous research showed that individuals who perceive COVID-19 as emotionally threatening, feel lonely or are less resilient exhibit higher mental distress.24

On the other hand, it is also likely that mental distress emerges as result of the damaging effects of distancing measures and stay-at-home orders on the economic situation of families.27 Indeed, economic crisis may affect mental health either by increasing risk factors (i.e., unemployment, indebtedness, loss of socioeconomic position) or by weakening protective factors (i.e., job security and welfare programs).24 Moreover, economic crisis has been associated to increased need for general help sought for mental health problems, higher use of prescription drugs and increased hospital admissions for mental disorders.24

Our findings and those of other studies suggest that women are in greater risk of suffering depression, as well as sleep and eating disorders in comparison to men.7,23,24,28,29 Higher vulnerability of women to experience mental distress is not well understood, but it probably relates to gender differences regarding access to health care and gender expected roles in society and economic production.30 In comparison to men, women generally have less access to health care and their diseases diagnosed later, which as a result cause them to be treated less successfully. Additionally, women usually have to balance work and household tasks being more vulnerable to suffer psychological burden.30 On the other hand, it is also know the influence of sex-based differences on certain mental health outcomes, which are not taken into account to provide medical treatment to women.30,31

Similarly to other studies, we also found an association between mental health distress and age during the COVID-19 pandemic.7,32,33 In general, it seems that youngsters had suffered higher mental health burden due to social isolation than did older people during confinement. In the context of the pandemic youngster's vulnerability might be explained by the abrupt and drastic change in their daily life-style. Sudden isolation and disintegration of their social networks could have prompted feelings of loneliness, which has been linked to mental health problems such as stress and anxiety in the future.34 Moreover, during the lock-down youngsters, who already are at higher risk of mental health problems and self-harm than other age groups, have been exposed to situations that exacerbate self-harm and suicide ideation, including feeling trapped, defeated and hopeless.34

The analyzes carried out in the population regarding health-disease phenomena derived from confinement require the attention of the scientific community for the formulation of adequate health policies. In this way, our findings highlight the need of promoting resilience specially among the most vulnerable population, i.e. women and youngsters in order to help them overcome the devastating consequences of the pandemic.35 Moreover, our study confirms the importance of discussing the implementation of gender-specific health measures to monitor and address the negative effects of the pandemic.

Our study is important because it is the first one conducted in Ecuador including near 9000 participants. Moreover, it provides for the first time an insight on the association of lock-down and mental health of young adults and differences among women and men. However, our study has several limitations: first, although we evaluated changes in mobility for the population in general, we failed to extract information on personal compliance to the mandatory confinement; second, our study lacks of external validity since participants came from a non-probabilistic sampling, and therefore our findings can not be generalized to the general population; and finally, due to our study design we were not able to establish a causal relationship between mobility restrictions, nor to address the impact of other important factors such as environmental context.