This cohort study was conducted in the Spanish region of Málaga (Andalucía) and approved by the North-East Malaga Ethics Committee (1078-N-20). Interviews were telephone-administered to guarantee the safest means to communicate during the COVID-19 pandemic. Researchers contacted participants by telephone, explained the study in detail, answered any questions that arose, and obtained consent from those willing to participate in the study [23].

The study was approved by the North-East Malaga Ethics Committee (1078-N-20). Participants provided written consent before taking part.

This study was registered in ClinicalTrials.gov (NCT04385797).

Participants were identified from the Support, Monitoring and Reminder Technology for Mild Dementia (SMART4MD; NCT03325699) [24] and TV-Based Assistive Integrated Service to Support European Adults Living with Dementia (TV-AssistDem; NCT03653234) [25] randomized clinical trials (RCTs), which aimed to assess the effects of ICTs to support MCI/MD using a tablet-based health application and a television-based assistive integrated service, respectively. In both RCTs, a broad definition of MCI, a subjective memory deterioration sustained over time, was considered. All participants had undergone 1-3 previous assessments (in 6-month intervals) in the RCTs on cognition, quality of life, and depression prior to the COVID-19 breakout.

Researchers from the Biomedical Research Institute of Malaga contacted 210 potential respondents from the SMART4MD (n=111, 52.9%) and TV-AssistDem (n=99, 47.1%) RCTs by telephone. In total, 151 participants, SMART4MD (n=75, 49.7%) and TV-AssistDem (n=76, 50.3%), agreed to participate. However, for 8 (5.3%) of them, it was not possible to assess the main variable (cognition) and the secondaries variables (quality of life, perceived health status, and depression), because their abilities to answer the questionnaires were compromised during the time of assessment.

Participants were eligible for inclusion when the following criteria applied: participating in the SMART4MD and TV-AssistDem RCTs and agreeing to participate by giving consent. Eligibility criteria of the aforementioned RCTs were age>55 years or >60 years, perception of memory problems for at least 6 months, score of 20-28 or 23-27 points in the Mini-Mental State Exam (MMSE), independently living, having an informal caregiver, and taking care of their medical prescription. Patients with a score above 11 on the Geriatric Depression Scale (GDS), a terminal illness, or specific cognitive or physical conditions that would reduce their ability to use a tablet or a television were excluded.

Participants were contacted by telephone by 5 health care professionals (2 neuropsychologists, 1 clinical psychologist, 1 psychologist, and 1 psychiatric and mental health clinical nurse specialist). Researchers had previously established relationships with participants during both RCTs. Quantitative and qualitative strategies were used to create an unstandardized ad hoc telephone-based survey in order to gather as much information as soon as possible. The exceptional situation did not allow us to test the instrument prior to its implementation by phone. To minimize the interference of this situation in the results, validated phone versions tests were used.

The survey was a useful tool for guiding the interviewers and gathering information simultaneously in a homogenous way. A model of the questionnaire used is attached in Annex 1 in Multimedia Appendix 1.

Researchers interviewed the participants between May 11 and June 26, 2020. The variables of sociodemographic data (age, sex, and living arrangements), health perception-management (change in living arrangements due to lockdown, presence of COVID-19 symptoms, frequency of access to COVID-19 information), sleep-rest patterns, types of ICTs (smartphone, tablet, television, laptop), and their uses (informative, cognitive, entertainment, and socialization) were collected from the participants unless their abilities to answer such a long interview were compromised, in which case the caregivers were interviewed on their behalf. The questionnaires that evaluated the main variables (cognition, quality of life, depression, perceived stress, and technophilia) were answered by the participants.

The mean time from the start of the lockdown and home confinement measures to the interview was 70.36 days (SD 12.40, range 52-102).

The flow of participants is shown schematically with counts in a participant flow diagram (Figure 1). Statistics considered for presentation for continuous measures were the mean and SD, and if the criterion of normality was not met, the median and the first and third quartiles. Categorical variables were summarized using counts and percentages.

The change in means in the main outcome (cognition) and in the secondary outcomes (quality of life, perceived health status, and depression) were analyzed with respect to the last assessment of the RCTs (SMART4MD and TV-AssistDem) using the repeated measure t test or the nonparametric Wilcoxon test, if appropriate (considering significant values of α<.05). In the secondary outcomes, we applied Bonferroni correction for 3 comparisons (considering significant values of α<.017).

For the analysis of groups based on living arrangements (living alone vs living with others), the t test was used for continuous variables and the chi-square test for categorical variables. To evaluate the association of this variable with mental health outcomes, multivariate logistic regression for binary variables or linear regression for continuous variables was performed to adjust for confounders: age, sex, and technophilia. In addition, we explored the association between a change in living arrangements during the pandemic and cognition, depression, quality of life, and perceived health status with univariate and multivariate linear regression (adjusting for the following confounders: sex, age, and current living arrangements). For this comparison, we applied Bonferroni correction for 4 comparisons (considering significant values of α<.013).

For the analysis using groups based on the score in technophilia (based on the median of the TechPH index as the cut-off point), the t test or nonparametric Wilcoxon test was used for continuous variables and the chi-square test for categorical variables. To evaluate the association of this variable with mental health outcomes, multivariate logistic regression for binary variables or linear regression for continuous variables was performed to adjust for confounders: age, sex, and living arrangements.

To establish the Bonferroni correction for multiple comparisons (regarding living arrangements and technophilia groups), the number of independent tests was estimated with principal component analysis. Of the 37 variables analyzed, the first 35 components explained >99% (99.3%) of the variance. Thus, we considered the values significant with α<.0014.

To analyze the assumptions of all linear regression models, the Ramsey RESET linearity test, the Breusch-Pagan homoscedasticity test, and the Shapiro-Wilk normality test of the model residuals were used (see Annex 2 in Multimedia Appendix 1). When cognition, quality of life, health status, and depression were used as dependent variables, the assumption of the normality of the residuals of the model was not fulfilled, and transformation did not solve the problem, robust linear regression models were used (using the robustbase package and the lmrob function in R).

The R version 4.0.4 program was used for all statistical analysis [41].

Of the 210 potential respondents (n=111 [52.9%] from SMART4MD and n=99 [47.1%] from TV-AssistDem), a total of 165 (78.6%) of 210 respondents was successfully reached, of which 151 (91.5%) agreed to participate (Figure 1). In addition, 150 (99.3%) participants completed the full interview without assistance, and 15 informal caregivers were interviewed on behalf of participants whose abilities to answer such a long interview were compromised. Of these 15 participants, 11 (73%) answered the main variables questionnaires and 4 (27%) were unable to do so. Given the complexity of MCI/MD, to not skew the results of patients with a greater decline, all data were considered.

The mean time between the last assessment of the RCTs (T0) and the interview during the lockdown (T1) was 199.33 days (SD 52.43, range 67-395), and the mean duration of the calls was 50.14 minutes (SD 16.40).

The mean age of the sample was 74.31 years (SD 6.48), 97 (64.2%) of 151 participants were women, 36 (23.8%) lived alone, and 80 (53.3%) had high attraction to technology (high technophilia). The COVID-19 outbreak forced 22 (14.6%) of 151 participants to change their living arrangements (Table 1).

Regarding the differences between the period before and during the outbreak, there were no differences in the main outcome: cognition. After correction for multiple comparisons, there were no statistically significant differences in the quality of life, perceived health status, or depression between the 2 periods (Table 2).

Regarding social isolation (living alone and living with others), after Bonferroni correction, there was no significant association between the variables of the study (Table 3). Some factors reached nominally significance: self-perceived fear (being alone 36.1% vs 18.4%; χ²=4.27; P=.04) and depression (being alone 3 vs 2; Z=–2.10; P=.04). In the multivariate models, depression did not reach nominally statistically significance (B=0.83; P=.06); see Table 3 (note: significant results with P<0.001). After Bonferroni correction, there were statistically significant differences regarding home visits, with those living alone (being alone 82.9% vs 51.3%; χ²=10.97; P<.001) receiving more visits than those living with others. There was no significant association regarding other variables (only a nominal significant association regarding the more frequent use of the newspaper in those living with others).

The change in living arrangements was not associated with cognition (unadjusted model: B=–0.21, P=.65; adjusted model: B=0.36, P=.44), quality of life (unadjusted model: B=–0.74, P=.40; adjusted model: B=1.05, P=.21), perceived health status (unadjusted model: B=3.34, P=.41; adjusted model: B=3.34, P=.41), or depression (unadjusted model: B=–0.15, P=.74; adjusted model: B=–0.25, P=.56). After correction for multiple comparisons, there was no association with less perceived stress (unadjusted model: B=–4.72, P=.01; adjusted model: B=–0.26, P=.02) but the results reached nominal significance. Annex 2 in Multimedia Appendix 1 shows more detailed information about linear regression models.

After correction for multiple comparisons, there was no significant association between technophilia and the variables of the study. Only some variables reached nominal significant associations: self-perceived boredom (high technophilia 10.1% vs 27.7%; χ²=7.44; P=.01), calmness (high technophilia 31.6% vs 52.3%; χ²=6.30; P=.01), perceived stress (high technophilia 18.1 vs 21.23; t=2.19; P=.03), depression (high technophilia 2 vs 3; Z=2.16; P=.03), and quality of life (high technophilia 37.3 vs 35.3; t=2.24; P=.03). In the multivariate models, after controlling for possible confounders, the associations only maintained nominally statistical significance and the perceived health status reached nominal significance (B=6.44, P=.04); see Table 4 (note: significant results with P<.001).

Of 148 participants, 39 (26.4%) reported accessing extreme and 32 (21.6%) reported accessing too much COVID-19 information. The most frequent ICT used to access COVID-19 information was mainly the television (134/147, 91.2%), and most participants were also informed through family and friends (120/148, 81.1%). Furthermore, only 46 (30.7%) of 150 participants did not contact health or social services (Table 5; note: significant results with P<.001).

Although most of the participants (46/151, 30.7%) preferred paper-based memory exercises, the most frequent ICT used for cognition was the television (16/151, 10.7%). The most used ICTs for entertainment were the television (138/151, 92%), followed by the smartphone (60/151, 40%), and for socialization, the smartphone (75/151, 50.3%). Detailed information is given in Table 6 (note: significant results with P<.001).

This cohort study was conducted to understand the impact of restrictive measures in community-dwelling older adults with MCI and MD during the first COVID-19 outbreak.

Our findings show that the first months of the outbreak did not significantly impact the cognition, quality of life, perceived health status, and depression of our study population when making comparisons with baseline assessments prior to the outbreak. Change in living arrangements had no influence on these variables either. Living alone and technophilia were not associated with mental health–related variables after correction for multiple comparisons. However, being alone was nominally associated with self-perceived fear and depression, and higher technophilia with better quality of life, less boredom, perceived stress, and depression but also less calmness. Overall, health care and social support service access and utilization were high. The most used ICTs during the COVID-19 outbreak were the television for informative-, cognitive-, and entertainment-related uses and the smartphone for socialization.

To the best of our knowledge, few studies have addressed the consequences of the COVID-19 outbreak on the cognition of the elderly, and the use of technologies during this ongoing societal change.

Several studies have shown that quarantine measures have changed the behaviors and lifestyle of older people with cognitive decline [42], although in some cases, these changes have been less important than expected [43]. Lifestyle changes can increase the risk of dementia and cause cognitive impairment. However, our study showed that in our sample, the first stages of the COVID-19 outbreak did not cause significant cognitive decline in comparison with a previous assessment. This result could be explained by 2 reasons. On the one hand, the evaluation was carried out on an average of 70 days after the start of the home confinement restrictions and, likely, this time was not enough to influence cognitive decline. On the other hand, the data of our study show that most of the participants maintained an active lifestyle and used new technologies for cognitive stimulation, information access, leisure, and social connectedness. This combination of healthy lifestyle factors and opportunities for cognitive and social stimulation has proved to be important in reducing the risk of cognitive decline [44].

Regarding the lack of differences in the quality of life and perceived health status before and during the outbreak in people with MCI/MD, a similar conclusion was reached by another cohort study in a similar population in Spain using the EuroQoL-5D-3L [45]. Other studies on quality of life in different population groups during the pandemic have also not found a perception of poor quality of life in the elderly [46]. Our results could be explained because our sample has continued to carry out cognitive, entertainment, and social activities, which are associated with the definition of a good quality of life by the elderly [47]. Another factor that can lead to a good quality of life is maintaining well-being and good sleep quality. Poor sleep quality is known to have a significant impact on lower levels of life satisfaction and mood [48].

Regarding mental health, longitudinal studies have established that it is has been affected by the pandemic [49-51], but frequently, the impact has been higher in the younger population and those more economically vulnerable [51,52]. Likewise, another longitudinal study in Spain with a sample of older adults with dementia or cognitive impairment found an increase in depressive and anxious symptoms after the confinement [53]. Furthermore, attention has been drawn to the possible harmful effects of the excessively dramatic presentation of the consequences of the restrictions due to the pandemic, which are based mainly on survey studies, in many cases carried out without the required rigor [54]. However, the results are mixed, and, for example, some longitudinal studies in Spain [55] and Greece [56] did not find differences regarding depression when comparing the period before and after confinement, and another Dutch community study did not find an increase in depressive symptoms in the general population [57]. In a cross-sectional study comparing populations over and under the age of 60 years during the peak of the COVID-19 pandemic in Spain, the elderly did not demonstrate special vulnerability to acute stress and no sex differences were found. This study hypothesized a greater resilience in the elderly due to the economic and social difficulties experienced throughout their lives during the Spanish post–Civil War period (1939-1960), increasing their ability to cope with stress and face the pandemic resiliently [52]. The results of this study in which no differences were found in depressive symptoms could also be explained by the specific characteristics of the study sample, which present MCI and mainly maintain autonomy to live independently. In addition, the evaluations were conducted when the stricter measures of the first lockdown in Spain were being brought down. Moreover, the small incidence of the COVID-19 virus at the time of assessment could have influenced the results (infections less than 1%). These results show the complexity of the effects of the pandemic, highlighting the need for more longitudinal studies in different populations to evaluate the effects of the social restrictions and the pandemic.

Another factor to consider was whether living alone during COVID-19 confinement was associated with a higher prevalence of depression. Although several studies have found a significant association between depression and living alone during the pandemic [42,58], others have shown otherwise [59,60]. Our results are in line with the second ones, showing no significant differences in GDS scores. However, more studies are necessary to determine whether living alone is a risk factor for depression during the pandemic. The results are in line with our previous findings [61] that associated living alone with worse mental health at the beginning of the outbreak (without correcting for multiple comparisons). The way our sample used the ICTs, through online communication, remote social interactions, or video calls, could have been useful to address social isolation during the pandemic [62].

Regarding technophilia, our study did not find an association between a better attitude toward technology and better mental health. In line with these results, a multicenter study conducted in Norway, the United Kingdom, the United States, and Australia also found no change in loneliness and the quality of life in adults over 70 years who used ICTs to maintain social contact during the COVID outbreak [63]. However, we showed in a previous study based on the TV-AssistDem RCT how technology could be useful to maintain cognitive activities [61], and more studies need to clarify whether the evidence supports the recommendations on interventions that may improve the knowledge of ICTs and are related with the use of technology to maintain social connections and cognitive activities [64].

A main limitation of this study was changing the interview administration from face-to-face before the outbreak to telephonic during the outbreak. Interviews were performed by the same professionals in both cases to reduce this possible bias, the measures were rescaled accordingly, and the validated phone versions of the tests were used. In addition, the interview had a mean duration of 50.14 minutes, which could cause fatigue in this population and alter their performance.

Another limitation is that the sample came from 2 RCTs and the participants who agreed to participate in the RCTs may have special characteristics that make them not representative of the general population. Furthermore, the sample was from only 1 center in Andalusia. However, it is a larger sample than in other studies carried out in Spain to date with this type of population.

Moreover, 15 caregivers answered on behalf of patients whose ability to answer for themselves was compromised. Although they did not respond to the questionnaires that evaluated the main variables, their answers may have interfered with the results.

The Bonferroni correction for multiple comparisons is conservative and could have increased type II errors.

Some studies have pointed out that the effects of changes during the lockdown may be temporary compared to long-lasting ones. Therefore, future effects will need to be explored as it is possible that once the lockdown is over, many people may not return to their “normal routine” as before the pandemic and will continue to avoid face-to-face activities, especially those regarding social and physical activity due to the fear of the contagion [42].

During the COVID-19 outbreak, governments’ restrictive measures demonstrated being effective in viral spread prevention. Although these restrictions have had negative effects on health and well-being and have changed lifestyles worldwide, our study showed how a presumably vulnerable population has shown more resilience to restrictive measures than expected. The people with MCI/MD did not experience a significant decline in cognition, quality of life, perceived health status, or depression during the period of the COVID-19 outbreak. The study also showed that being alone and a negative attitude toward technology are not associated with worse mental health after correcting for multiple comparisons. In addition, the data were collected over a short period, and further research is needed to explore whether maintaining restrictive measures for longer influences a worsening of cognitive abilities, quality of life, perceived health status, or depression and which factors increase the risk of poor mental health in this population. They reported overall well-being, maintained sleep quality, and presented moderate perceived stress. This could be related to the fact that our sample continued participating in daily activities, which plays a crucial role in enhancing and maintaining cognition [65,66], just like keeping social interaction. The use of ICTs was essential to carry out these activities during the restrictions. Television was the most widely used ICT for informational-, cognitive-, and entertainment-related uses, and the smartphone for socialization.