Abstract
Background:
Pandemic fatigue generates low motivation or the ability to comply with protective behaviors to mitigate the spread of COVID-19.
Aims:
This study aimed to analyze the symptoms of pandemic fatigue through network analysis in individuals from five South American countries.
Method:
A total of 1,444 individuals from Argentina, Bolivia, Paraguay, Peru, and Uruguay participated and were evaluated using the Pandemic Fatigue Scale. The networks were estimated using the ggmModSelect estimation method and a polychoric correlation matrix was used. Stability assessment of the five networks was performed using the nonparametric resampling method based on the case bootstrap type. For the estimation of network centrality, a metric based on node strength was used, whereas network comparison was performed using a permutation-based approach.
Results:
The results showed that the relationships between pandemic fatigue symptoms were strongest in the demotivation dimension. Variability in the centrality of pandemic fatigue symptoms was observed among participating countries. Finally, symptom networks were invariant and almost identical across participating countries.
Conclusions:
This study is the first to provide information on how pandemic fatigue symptoms were related during the COVID-19 pandemic.
Introduction
The negative impact of the COVID-19 pandemic on mental health worldwide has been well documented (Haktanir et al., 2022; Torales et al., 2021). Individuals have been reported to experience increased levels of stress, anxiety, depression, hopelessness, and sleep problems, among others (Clemente-Suárez et al., 2021; Jahrami et al., 2021; Kontoangelos et al., 2020; Stavridou et al., 2020; Xiong et al., 2020). In addition, deterioration in family dynamics and daily lifestyle has also been reported (Fountoulakis et al., 2022). This reality is not foreign to Latin America, where high but heterogeneous levels of mental health symptom prevalence have been observed during the pandemic (Gallegos et al., 2022; Zhang & Chen, 2021; Zhang et al., 2022). In South America, the impact of COVID-19 on mental health was greater in low-income populations (Bassey et al., 2022). Likewise, mental and physical exhaustion has been reported as a consequence of COVID-19 restrictions (da Silva, 2021). This phenomenon has received numerous designations, such as ‘quarantine fatigue’ (Marcus, 2020), ‘emergency fatigue’ (Michie et al., 2020), ‘pandemic exhaustion’ (Queen & Harding, 2020), and ‘pandemic fatigue’ (Murphy, 2020; World Health Organization [WHO], 2020). However, it has been suggested that the concept of ‘pandemic fatigue’ may be the most appropriate (Haktanir et al., 2022), as it refers to psychological fatigue involving feelings of physical and mental tiredness and exhaustion (Michie et al., 2020). As a result, pandemic fatigue generates low motivation or the ability to comply with protective behaviors to mitigate the spread of COVID-19 (Michie et al., 2020). Similarly, the WHO (2020) indicates that pandemic fatigue is the lack of motivation to perform protective behaviors and seek information about COVID-19.
It has been suggested that pandemic fatigue is a natural response to a prolonged public health crisis owing to its relationship with social isolation (Zerbe, 2020), the large amount of news about COVID-19 in different information media (Teng et al., 2020; Torales et al., 2022), and self-care (Zou et al., 2020). Moreover, from a physiological point of view, the development of pandemic fatigue can be explained by human physiology. In this regard, it has been suggested that the release of adrenaline in the early months of the pandemic generated enthusiasm for coping with the pandemic. However, it is difficult for the body to maintain adrenaline doses for a long period of time, even more so if the development and end of the pandemic were uncertain in the first months (Murphy, 2020). This caused the initial enthusiasm to wane and generated general exhaustion in the face of the pandemic (Murphy, 2020). Pandemic fatigue can be expressed in different ways throughout the course of a pandemic (Haktanir et al., 2022). Thus, individuals may gradually become irritated with the norms related to COVID-19, believing that the norms restrict their freedom, consider COVID-19 to be less dangerous than it was initially, feel pessimistic and hopeless about the effectiveness of methods to control the pandemic, or feel an increased desire to socialize with others (Harvey, 2020; Taylor et al., 2022). In addition to the fact that pandemic fatigue decreases motivation to cope with the pandemic and complies with the restrictive measures given by different governments (WHO, 2020), it has also been related to the presence of sleeping difficulties, fear, and worry (Labrague & Ballad, 2021), and negatively impacts well-being (WHO, 2020). Similarly, pandemic fatigue has a major impact on adherence to vaccination programs and how people cope with infectious diseases (Taylor et al., 2022). Factors associated with the emergence of pandemic fatigue include age, gender, educational level, ethnicity to which one belongs (MacIntyre et al., 2021), and perceived risk of disease (Cleofas & Oducado, 2021, 2022; Wright et al., 2022). Regarding the latter, a decreased perceived risk of COVID-19 is associated with an increased level of fatigue (Buttenschøn et al., 2022; So et al., 2017).
Theoretically, pandemic fatigue is defined based on two dimensions: lack of motivation to continue performing recommended protective behaviors, and boredom due to the excessive amount of information about the pandemic (Cuadrado et al., 2021). The presence of demotivation is directly related to the definition provided by WHO (2020) and other authors (Hassan et al., 2021). In the case of boredom, it has been shown that overexposure to general or specific information about COVID-19 increased risk perception and panic during the pandemic (Liu et al., 2020). These two dimensions suggest that pandemic fatigue is a negative attitude toward pandemic protective behaviors that generates emotional exhaustion related to the increased harshness of constraints and overexposure to COVID-19 information (Murphy, 2020; Reicher & Drury, 2021).
The existence of pandemic fatigue has been a subject of debate. Some have denied its presence because of limited evidence on the true impact it may have had on decreasing adherence to health protection measures (Harvey, 2020; Michie et al., 2020). However, there are also those who consider pandemic fatigue to be a real problem based on behavioral observations (Franzen & Wöhner, 2021). As already mentioned, pandemic fatigue has affected people’s behavior and, therefore, affected the evolution of the COVID-19 pandemic. Therefore, pandemic fatigue has become an important factor in the spread of COVID-19. Pandemic fatigue has been examined extensively based on traditional approaches that use total scale scores to describe symptom severity and its relationships with other mental health variables (Labrague, 2021; MacIntyre et al., 2021; Zarowsky & Rashid, 2022). However, these approaches may miss meaningful relationship between individual symptoms (Cai et al., 2022). To address the limitations of the traditional approach, network analysis has been proposed as a more recent approach to better understand the co-occurrence of mental health symptoms (Borsboom & Cramer, 2013).
Traditionally, latent illnesses such as depression, anxiety, or fatigue are considered to produce symptoms such as problems with sleep, concentration, and mood. However, network analysis points out that symptoms of mental health problems do not arise from underlying factors but interact and reinforce or inhibit each other (Borsboom, 2017). In this sense, there are symptoms that have a greater influence on the network, called core symptoms, which generate the emergence of other symptoms (Beard et al., 2016; Jones et al., 2021; McNally et al., 2015). That is, nodes are assumed to cluster, as they somehow have a relationship with each other. As nodes interact directly with each other, there is a higher probability that variations in a central node will generate changes in other nodes in the network (Epskamp et al., 2018). Identifying core symptoms allows not only obtaining information about symptoms that are highly related to others but also allows the exclusion of symptoms that are not related to others (Elliott et al., 2020). The links between the nodes are called edges. Likewise, one or more symptoms of a mental disorder may also trigger symptoms of another disorder, leading to the emergence and maintenance of comorbidities (Cramer et al., 2010). The symptoms that link disorders are referred to as bridging symptoms (Kaiser et al. 2021). Network analysis allows the identification of core and bridging symptoms within the network for a better understanding of the symptoms that put people at greater risk for comorbidity in order to develop more effective interventions (Borsboom & Cramer, 2013; Fried & Cramer, 2017).
A symptom network is not a static but a dynamic process (Forbes et al., 2017; Robinaugh et al., 2020), and assessing changes in symptoms within different countries can help gain a deeper understanding of the mechanisms behind the persistence of comorbidity (Bringmann et al., 2016). The different COVID-19 prevention policies implemented by countries did not necessarily express cultural differences, but they did impact the mental health of citizens, how the risk of COVID-19 was assessed, and how its consequences were interpreted (Lin et al., 2021). This is even more important in Latin American countries where the presence of precarious health systems, poor health infrastructure, governance problems, higher prevalence of chronic diseases, inequity, and a large percentage of people living in poverty have contributed to the severe impact of COVID-19 (Meda-Lara et al., 2021; Pablos-Méndez et al., 2020). Different cultural, social, and governmental systems have been suggested to affect the gradual development of pandemic fatigue (Haktanir et al., 2022). A recent study reported the presence of average levels of pandemic fatigue among individuals from South American countries, where Peru presented the highest level of pandemic fatigue, followed by Bolivia, Argentina, Uruguay, and Paraguay (Torales et al., 2023).
Considering the above, the present study aimed to analyze the symptoms of pandemic fatigue using network analysis in individuals from five South American countries. In addition, we had the following specific objectives: (a) to analyze the network structure of pandemic fatigue symptoms, (b) to identify the most important symptoms using centrality and predictability indices, and (c) to compare the networks among the South American countries evaluated.
Methods
Participants
The present study was conducted on 1,444 individuals from Argentina, Bolivia, Paraguay, Peru, and Uruguay. In each of these countries, there were researchers interested in participating. However, the inclusion of more Latin American countries could help broaden international comparisons of the networks. Participants were selected according to the following inclusion criteria: being of legal age according to the legislation of each country, residing in one of the five countries mentioned, and providing informed consent to participate in the study. The snowball technique was used as the sampling method to obtain a significant number of responses (Roy et al., 2020). However, using this method does not allow us to affirm that the samples are representative of each country. The sample size was previously defined using the Monte Carlo simulation method (Constantin et al., 2021), which recommended a minimum number of 710 participants. Finally, the number of participants included in the study far exceeded the suggested number.
It was observed that in all five countries, the sample was mostly female (> 68% in all countries), with a lower percentage of males, and very few individuals reported being of non-binary sex or preferred not to indicate their sex. Their average age ranged from 33.05 years (Uruguay) to 35.01 years (Paraguay). Regarding marital status, most participants were married or had a partner (42.71% vs. 47.02%). Finally, the highest percentage of participants reported having a university education (>88%). The fact that the samples in each country are made up mostly of women, adults, married or with a partner and with a university education, also limits the generalization of results to the population of each country. More specifically, Table 1 shows the sociodemographic characteristics of the participants in each country included in the study.
Participants’ sociodemographic characteristics.
Measures
The Pandemic Fatigue Scale (PFS; Cuadrado et al., 2021), originally developed in Spain, assesses pandemic fatigue, which is a natural response to the repetitive execution of protective measures to mitigate the spread of COVID-19 and to pandemic information overload. The PFS consists of six items, three of which assess the dimension of carelessness or lack of motivation to follow protective measures (items 1, 2, and 3), while the other three items (items 4, 5, and 6) assess the dimension of boredom with information about COVID-19. All PFS items have seven Likert-type response options (1 = ‘strongly disagree’ to 7 = ‘strongly agree’). The sum of the scores for each item gives the total score for each dimension. Higher scores indicate greater symptoms of pandemic fatigue. In this study, we used a previously used Spanish version (Torales et al., 2023), which has been shown to have an invariant structure and adequate reliability in a set of Latin American countries (Caycho-Rodríguez et al., 2023).
Procedure
An online survey was developed using Google Forms, which consisted of a section on sociodemographic data and another section on PFS. Using online surveys allows for greater global reach and faster collection of information, especially in emergency situations, such as COVID-19. However, among the limitations of the use of online surveys, we can mention the little information about the population that had access to the survey, and about who actually responded to it. Furthermore, using an online survey had the limitation of leaving those who did not have access to the Internet out of the sample. This makes it difficult to define or describe the population to which the results can be generalized (Andrade, 2020). Prior to the questions, the survey presented the objectives of the study and informed consent was obtained. The online survey was distributed to individuals who met the inclusion criteria through social networks (Instagram, Facebook, and Twitter) and mailing lists between November 1 and December 20, 2022. This time period corresponds to the final stage of the COVID-19 pandemic, so the findings express only experienced in this period. The average response time to the online survey questions was approximately 10 minutes. The data collection procedure was the same for all participating countries.
Ethical considerations
The present study was approved by the Department of Medical Psychology of the National University of Asunción (Paraguay; approval number 53/2022). Adherence to the principles of confidentiality, equality, and justice as outlined in the Helsinki Declaration were strictly maintained throughout the data collection and analysis process. All participants provided informed consent and were informed that they could withdraw from the study at any time.
Data analysis
In this study, the R programming language in the RStudio environment (RStudio Team, 2022) was used to perform all analyses. The ‘qgraph’ 1.9 (Epskamp et al., 2012) and ‘bootnet’ 1.5 (Epskamp, 2020) libraries were used to create the network plots. For multi-sample estimation, the recommendations provided by Fried et al. (2018) were followed, which include network estimation, network stability, network inference, and network comparison.
Network estimation
The networks were estimated independently using the ggmModSelect estimation method and a polychoric correlation matrix (Isvoranu & Epskamp, 2021) was used. A spin-glass algorithm was used to obtain a measure of node communalities.
Network stability
The stability of the five networks was evaluated using the bootnet 1.5 package (Epskamp, 2020). For this purpose, a nonparametric resampling method based on the case bootstrap type, consisting of 1,000 replications, was used. The correlation stability coefficient (CS) was used, which indicates the maximum number of cases that can be removed with a probability of 95% to retain a correlation of at least 0.70 between the original network-based statistic and the statistic calculated with fewer cases. The CS value should not be less than 0.25 and it is recommended that it be greater than 0.50.
Network inference
In this study, we estimated network centrality using a metric based on node strength, which takes into account the absolute values of the edges in order to identify the most influential node. It is worth noting that such a metric has been shown to be extremely stable in the field of psychology, as evidenced in previous research (Hallquist et al., 2021). The mgm 1.2-11 package (Haslbeck & Waldorp, 2020) was used to assess the predictive ability of the nodes. Owing to the ordinal nature of the data, the normalized predictability accuracy (‘nCC’) was used, which indicates the extent to which a node can predict all its neighboring nodes (Haslbeck & Waldorp, 2018). This information can be observed in the black circle surrounding each node in the graph.
Network comparison
In the present study, the NetworkComparisonTest 2.2.1 (NTC) library (van Borkulo et al., 2022) was employed to perform a network comparison. We set a fixed seed in 2023 and conducted an omnibus test to determine whether all network edges were identical. A permutation-based approach was used to conduct this test, using a two-tailed test in which the difference between the two groups was calculated by performing 1,000 replicates for each individual that was randomly regrouped. If the significance was less than .05, the null hypothesis that both networks were equal was accepted.
Results
Descriptive statistics
Figure 1 shows the response rates for each item organized by country. It can be seen that responses tended to be concentrated in the central values of the Likert scale (2, 3, and 4), suggesting moderate opinions of respondents in these countries. The highest concentrations of responses were generally found at a value of 4 for most of the factors assessed (PF1 to PF6). This common pattern suggests that perceptions in these countries about pandemic fatigue tend to be similar rather than extreme, in terms of disagreement or total agreement.
Response rates for both tests.
Network estimation
Analysis of the adjacent country matrices showed differences and similarities in perceptions of Boredom (PF1, PF2, and PF3) and Neglect (PF4, PF5, and PF6). In Boredom, the highest values were generally between PF2 and PF3, with Peru showing the highest value (0.4723156), and Uruguay the lowest (0.4496751). In Neglect, the highest values were between PF5 and PF6, with Bolivia having the highest value (0.4600856) and Argentina having the lowest (0.4313712).
As for the lowest values in Boredom, Argentina and Uruguay had the lowest ratio between PF1 and PF4, while Bolivia, Paraguay, and Peru had the lowest ratio between PF1 and PF6. It is interesting to note that, in all countries, the strongest relationships were found in the Neglect dimensions, which could suggest a greater concern for this area in the population of these countries. This analysis provides valuable insights into the perceptions and attitudes regarding pandemic fatigue in different national contexts.
In the density analysis, it was observed that in four of them, 8 out of 15 edges (connections) were non-zero, representing 53.33% density in each case. Only in Uruguay, 9 out of 15 edges were nonzero, reaching a density of 60 %.
In relation to the centrality indices, it was observed that in all countries, there were positive and negative values for the different FPPF variables. A common pattern was that the highest values tended to be positive and the lowest tended to be negative. The largest value corresponded to Peru in PF3, whereas the smallest value was found in Paraguay in PF1. These results suggest variability in the responses between countries and the different PF variables analyzed. However, PF4 was central in Argentina, Bolivia, and Uruguay, whereas PF2 and PF3 were central in Paraguay and Peru, respectively (Figure 2).
Networks according to the studied countries.
Stability of the network
In terms of stability, the CS values were analyzed considering that they should not be lower than 0.25, and it is recommended that they be higher than 0.50. Argentina and Uruguay reported CS values equal to 0.36 and 0.25, respectively. Both countries presented CS values higher than the recommended limit of 0.25. However, these values indicated that moderate stability was shown in Argentina and a minimum level of stability was reached in Uruguay. In contrast, Bolivia (0.13), Paraguay (0.05), and Peru (0.21) presented values below the recommended limit of 0.25, indicating lower stability in these countries in relation to the Strength measure (Figure 3).
Stability of the five countries’ networks considering centrality indices.
Network inference
Figure 4 shows that a recurring pattern in all countries was the presence of minimal discrepancies between the Bootstrap mean (black line) and the Sample (red line), which denoted precision in the estimates. The largest differences for each nation were as follows: in Argentina, PF2 to PF4; in Bolivia, PF3 to PF6; in Paraguay, PF1 to PF5; in Peru, PF1 to PF4; and in Uruguay, PF2 to PF4. Additionally, reduced 95% bootstrap Confidence Intervals were observed in all five countries, implying that the variations between resamples were not significant.
Accuracy of the networks of the five countries.
Predictability analysis revealed that the PF1 item exhibited the lowest values in all countries, while the maximum values fluctuated between the different symptoms. For example, in Argentina, symptom PF6 had the highest predictability value; in Bolivia, symptom PF5; in Paraguay, symptom PF3; and in Peru and Uruguay, symptom PF2.
Comparison of networks
Table 2 provides a detailed comparison of the correlation and significance values for the pairs of networks between five South American nations, highlighting the relationship between their respective networks using Spearman’s Coefficient, the Network Invariance Test, and the Global Strength. The Spearman Coefficient shows high values, between 0.68 and 0.92, indicating a positive and remarkable correlation between the countries. As for the Network Invariance Test, we observed p-values ranging from .02 to .88. A p-value of less than .05, such as that found between Bolivia and Argentina (.02), suggests statistically significant differences in the network structures, implying that they are not identical. On the other hand, p-values such as .88 between Bolivia and Uruguay indicate that there is not enough evidence to claim that the network structures differ significantly. Regarding Global Strength, p-values vary between .38 and .98. Values close to 1 suggest a high similarity in the overall connection strength between the networks of the compared countries, as observed in the Uruguay–Paraguay pair (0.98), while lower values, although not lower than .05, could indicate variations in connection strength, but not strong enough to be considered statistically significant under the conventional threshold.
Correlation and significance values of the compared pairs of networks.
Note. 1 = Argentina; 2 = Bolivia; 3 = Paraguay; 4 = Peru; 5 = Uruguay.
Discussion
Various psychological factors have played an important role during the COVID-19 pandemic, including pandemic fatigue (Haktanir et al., 2022, Morgul et al., 2021). However, as mentioned earlier, the study of pandemic fatigue was conducted based on traditional approaches that rely on total scale scores to determine the severity of symptoms and their relationship with other mental health problems (MacIntyre et al., 2021; Zarowsky & Rashid, 2022). This may not account for the presence of significant relationships between the individual symptoms of pandemic fatigue (Cai et al., 2022). Therefore, the present study is, to the best of our knowledge, the first network analysis of pandemic fatigue symptoms conducted to determine how they are interrelated in samples from five South American countries.
First, the descriptive analysis of the responses of the participants from each country indicated the presence of moderate levels of symptoms in the dimensions of pandemic fatigue. These moderate levels are similar to those reported in the demotivation and boredom dimensions of pandemic fatigue found in a previous study involving a group of Latin American countries (Torales et al., 2023). The presence of moderate levels of pandemic fatigue can be explained by the fact that the data were collected at the end of the second semester of 2022, when restrictions were relaxed and the amount of information on the pandemic decreased considerably in all countries.
The network theory in mental health emphasizes the interconnectivity of symptoms. Regarding the relationships between fatigue symptoms, these are stronger in the demotivation dimension, which suggests a greater concern for this dimension in the samples from the participating countries. This was expected because, as stated, demotivation to follow COVID-19 protection recommendations was an expected and natural response to a sustained stressor, such as the pandemic, among the general population of the countries (Masten & Cicchetti, 2016; Torales et al., 2023). It has been suggested that at the onset of the pandemic, individuals in the countries participating in this study drew on their coping skills, aimed at short-term stress management; however, as the pandemic progressed, different coping styles triggered fatigue and demotivation (Torales et al., 2023). One factor that can lead to demotivation to follow recommended health behaviors is active participation in social networks (Chen et al., 2023). For example, in Peru, information overload on social networks positively impacts the generation of technostress, and the latter influences the development of burnout (Alvarez-Risco et al., 2021). In Paraguay, the dissemination of excessive information, often false, generated risky healthcare practices (Moreno-Fleitas, 2020) and depressive symptoms (Torales et al., 2022). In the Bolivian case, at the beginning of the pandemic, excessive information about COVID-19 on social networks was related to a greater perception of risk, which, in turn, was associated with the adoption of preventive behaviors (Zeballos Rivas et al., 2021). However, as the findings of the present study indicate, it seems that greater exposure to this type of information over time produced demotivation to follow these recommendations.
Variability was observed in the centrality of pandemic fatigue symptoms among participating countries. In the case of Argentina, Bolivia, and Uruguay, the most central or important symptom was ‘I am already so tired of the COVID issue that I am no longer as careful as I was at the beginning’, which refers to the fatigue of information about COVID-19, which impacts people’s health care. In the case of Paraguay, the most central symptom was ‘When someone starts talking about COVID, I am disinterested’; while, in Peru the most central symptom was ‘I don’t want to hear about COVID anymore’. Both symptoms refer to disinterest in COVID-19 and belong to the boredom dimension. Exposure to prolonged confinement, as happened during the pandemic, made people more likely to lose interest (Le & Nguyen, 2021). These three symptoms contributed most strongly to the overall network of pandemic fatigue symptoms in each country assessed. Each symptom is related to overexposure to the pandemic information. The severity of COVID-19 and its rapid evolution provoked an intense and incessant bombardment of pandemic information everywhere, inevitably generating an increased risk of mental fatigue and desensitization to information and protective behaviors after an initial period of anxiety and fear (Koh et al., 2020). That is, the initial enthusiasm for coping with the health crisis was replaced by feelings of exhaustion (Murphy, 2020). Information overload generates a reduction in people’s processing capacity, which is related to decision making regarding whether to follow preventive practices in the face of diseases (So & Popova, 2018). These results may lead to the hypothesis that interventions that aim to decrease exhaustion from pandemic information have the potential to alleviate other symptoms of the pandemic fatigue symptom network. In this sense, the goal is to design successful communications over a long period of time without message fatigue and encompass effective communication methods, such as the presence of narrative or non-narrative approaches (Koh et al., 2020).
In our study, networks were found to be invariant and almost identical among the participating countries. Although the different cultural, social, and governmental systems of the countries affected the gradual development of pandemic fatigue (Haktanir et al., 2022), it seems that in our study, the networks did not vary and the matrices were similar. This finding would allow for a comparison of pandemic fatigue symptom networks between countries. When facing new diseases, epidemics, or pandemics, the impact of the culture of the country on the consequences of health communication should be considered (Kahissay et al., 2017). This may result in people experiencing different degrees of information overload and fatigue due to the large amount of repeated preventive information against COVID-19. Thus, having information about possible differences or similarities in pandemic fatigue symptom networks among Argentines, Bolivians, Paraguayans, Peruvians, and Uruguayans may serve to develop health communication practices and strategies to disseminate prevention messages across these countries (Jia et al., 2022). This is even more important, as having evidence that symptom networks are invariant means that possible differences in symptom relationships can be attributed to genuine differences between the countries assessed, and not to measurement errors due to response biases or cultural differences in symptom understanding. Investigating the comparability of symptom relationships during pandemic fatigue is a fairly new practice. This study is one of the first to theorize cultural differences in pandemic fatigue symptoms.
Our study has limitations that offer opportunities for improvement in future research. First, participants from the five countries were recruited through snowball sampling. Therefore, it cannot be stated that the samples are representative of each country. This resulted in the samples in each country being mostly female, adult, married or with a partner, and university-educated. This limits the generalizability of our results to the population of each country. In this sense, future studies should replicate the research using sampling strategies that allow more representative samples to be obtained, controlling for distributions of gender, age, marital status, education, economic status, and other variables in different populations. Second, only five South American countries participated, where researchers from each country had a potential interest in participating and met the requirements of the study. However, international comparisons of these networks could be expanded with the inclusion of more Latin American countries and other cultural groups. Third, the study had a cross-sectional design; therefore, in these cases, network analysis suggests but does not establish causality. The most significant edges may indicate causal influences; however, studies with experimental designs are needed to establish causality. Thus, the results of network analysis provide a source of hypotheses about causal relationships between symptoms in a network (Taylor et al., 2020). It would be important that future studies use longitudinal designs where longitudinal methods of network analysis are tested (Martín-Brufau et al., 2020; Snijders, 2012). Longitudinal studies can effectively assess the possible causal relationships between the symptoms of pandemic fatigue. Fourth, data collection was conducted online, leaving those without internet access out of the sample. Future studies should use other data-collection strategies to access more sectors of the population. Fifth, the study was conducted in the final stages of the COVID-19 pandemic, so the findings express only this period. Sixth, pandemic fatigue was assessed using a self-report measure, which could generate social desirability bias. Future studies assessing the impact of a pandemic on mental health symptoms should consider using qualitative techniques such as interviews and focus groups to obtain information.
In conclusion, the relationship between pandemic fatigue symptoms was strongest in the demotivation dimension. Variability in the centrality of pandemic fatigue symptoms was observed among participating countries. Thus, in Argentina, Bolivia, and Uruguay, the most central or important symptom was ‘I am already so tired of the COVID issue that I am no longer as careful as I was at the beginning’; in Paraguay, the most central symptom was ‘When someone starts talking about COVID, I am disinterested’; while in Peru, the most central symptom was ‘I don’t want to hear about COVID anymore’. Finally, symptom networks were invariant and almost identical among the participating countries. The limitations mean that the results of the study should be interpreted with caution and call for further research on cross-cultural comparison of networks related to mental health symptoms. However, the study can be considered pioneering, as it was the first to provide information on how pandemic fatigue symptoms were related during the COVID-19 pandemic. The results of this study have direct theoretical applications to the measurement of pandemic fatigue symptoms. For example, pandemic fatigue has been considered a variable related to individual differences; however, the present study suggests that cultural differences may exist. The results initially suggest how cultural differences influence the different relationships between symptoms and the identification of different core symptoms between countries. Regarding practical implications, mental health professionals should consider identifying and intervening in the most central or important symptoms in each country, which will have a greater impact on other symptoms.
Footnotes
Acknowledgements
None.
Author contributions
TC-R and JT provided initial conception, organization, and main writing of the text. JV-L analyzed the data and prepared all figures and tables. IB, MW-C, AT-L, LV, LWV, and AM-d-C-T were involved in data collection and acted as consultants and contributors to research design, data analysis, and text writing. The first draft of the manuscript was written by TC-R, and all authors commented on previous versions of the manuscript. All authors have read, reviewed, and approved the final text of the article.
Conflict of interest
The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.
Data availability statement
The database is available with a request to the corresponding author.
Consent for publication
Does not apply.
Ethics approval
The present study was approved by the Department of Medical Psychology of the National University of Asunción (Paraguay; approval number 53/2022).
Funding
The author(s) received no financial support for the research, authorship, and/or publication of this article.
Informed consent statement
Informed consent was obtained from all individual participants included in the study.
