Introduction

Anxiety Disorders (ADs) are highly prevalent, affecting approximately 7.3% of the global population (Baxter et al., 2012). AD’s include conditions such as Generalized Anxiety Disorder (GAD), Panic Disorder (PD), Social Anxiety Disorder (SAD) and Specific Phobias. They represent the most common psychiatric disorders worldwide and are a significant contributor to disability, ranking sixth across the lifespan (Yang et al., 2021). ADs are also associated with a 7% increase in mortality due to suicide (Baxter et al., 2012). Moreover, the economic burden of ADs is substantial, impacting individuals living with these disorders as well as healthcare services (Kessler & Greenberg, 2002). As chronic conditions, ADs tend to persist throughout a person’s life, and their prevalence increases over time. Furthermore, they often co-occur with other psychiatric and physical comorbidities, exacerbating the costs associated with mental health treatment (Kessler & Greenberg, 2002).

AD’s can also lead to impairments in both work and social performance. People with AD’s tend to present higher rates of work absenteeism, decreased work productivity, difficulties resolving work related tasks, and unemployment (el-Guebaly et al., 2007; Plaisier et al., 2010). Social impairment in people with AD’s can be seen in terms of reduced social contacts, activities and supports, and perceptions of social disability, loneliness and reduced connection with others (Saris et al., 2017). These impairments in psychosocial domains can be found across different AD’s such as GAD, SAD and PD (Naragon-Gainey et al., 2014).

Impairments associated with anxiety are not exclusive to people who have been diagnosed with AD’s, but also affect those who present subthreshold levels of anxiety (Karsten et al., 2013). Indeed, subthreshold GAD, in which individuals experience mild but recurrent anxiety symptomatology, is twice as prevalent as GAD diagnosis (Haller et al., 2014). Prospective studies have evidenced that symptoms of anxiety, even if not severe enough to reach thresholds for an AD, are persistent and can often predict the development of later AD’s (Bosman et al., 2019; Karsten et al., 2011). Given the prevalence, impairment, and potential for later development of AD’s, studies focusing of individuals with subthreshold anxious symptomatology are essential.

The COVID-19 pandemic resulted in a significant increase in anxious symptomatology and in the presence of AD’s worldwide. Meta-analysis and systematic reviews focusing on symptoms’ increase during the pandemic estimate a 25,6% rise in cases of anxious symptomatology, around three times higher than previous estimates (Santabárbara et al., 2021; Santomauro et al., 2021). This global increase of anxious symptomatology further highlights the need for developing studies working from the perspective of identifying the factors associated with the manifestation of symptoms.

The implications that come with having recurrent anxious symptomatology has led anxiety research to be focused on possible variables associated with such distress. These symptoms have often been studied in relation to physical health and the experience of threatening life events (Garnefski et al., 2001; Mehta et al., 2007; Paukert et al., 2010).

The amount of threatening or stressful events experienced throughout the lifespan is positively linked to anxiety levels (Beurs et al., 2001; Tiguman et al., 2022). Specifically, elevated levels of anxiety have been associated to experiences of abuse, interpersonal problems, financial problems, unemployment, and illness or death of someone close (Beurs et al., 2001; van Veen et al., 2013; Schneider et al., 2021). In addition, in older adults and in youth, chronic health conditions (Himmelfarb & Murrell, 1984; Jones et al., 2017), and health-related quality of life have been found to be positively and negatively (respectively) correlated to the severity of anxiety symptoms (Liao et al., 2021; Strine et al., 2005). Although the association between perceived health, stressful events and anxiety seems to be consistent, studies have mainly focused on youth or older populations, hence, studies focusing specifically on these variables on the general adult populations are less frequent and would require further research.

There has been less focus on the cognitive variables associated with anxiety symptoms, although, executive functions and social cognition have been partially studied (Green et al., 2008; Hollocks et al., 2014). As executive functions are required to self-regulate during goal-oriented activities (Nyberg et al., 2021), they can affect a person’s abilities to cope with anxiety generating situations and can therefore be associated with the impairments mentioned above (Snyder et al., 2015). Several studies have found lower performance in executive functions to be associated with higher anxiety levels in populations such as primary care patients, older women, and neurodiverse individuals (Hollocks et al., 2014; Nyberg et al., 2021; Parra-Díaz et al., 2021; Predescu et al., 2020; Rozzini et al., 2009). Nonetheless, very few studies have focused on the relationship between executive functions and anxiety symptoms in the general adult population (Ajilchi & Nejati, 2017).

Similarly, social cognition domains could be relevant to understand the social impairments associated with anxiety symptoms (Plana et al., 2013). Nevertheless, this link has been rarely studied directly, so difficulties in social cognition in individuals with subthreshold anxiety can only be theorized. Studies associating anxiety to social cognition have generally focused on specific AD’s, mainly Social Anxiety, with mixed evidence supporting this link (Alvi et al., 2022; Hollocks et al., 2014; Mavrogiorgou et al., 2016; Pearcey et al., 2021). Very few studies have focused on other AD’s (Plana et al., 2013), while no studies have focused on subthreshold levels of anxiety.

Considering this background, the purpose of this research was to identify how relevant variables such as health indices, stressful life events, executive functions and social cognition are associated with the severity of anxiety symptoms. We chose to explore this set of multidimensional variables because they have not been previously explored together. To this end, we conducted two studies. The first one focused on identifying the association between anxiety symptoms and health indices and stressful life events, and whether this association varies between participants with no anxiety, mild anxiety, and moderate to severe anxiety. In the second one, we investigated the relationship between anxiety symptoms and executive functions and social cognition, as well as variations according to anxiety levels in a subsample of the first study. The results obtained from both studies contribute to a better understanding of the relationships between these variables and shed light on the most relevant factors for the prevention and treatment of subthreshold anxiety.

Method

Participants

The sample for Study 1 consisted of 548 Colombian adults, mainly residing in Bogotá (66.2% female, 33.8% male), with a mean age of 26 years (SD = 9.31), and 15.11 years of formal education (SD = 2.88). Only 16% of participants had a previous mental health diagnosis, with only 3% having been diagnosed with an AD. Participants were mostly students and graduates from different universities in Bogotá, who were invited to participate through email and social networks.

We evaluated participants in terms of their anxiety symptoms, their physical and mental health indices, and the amount of threatening life events experienced. We administered the scales through a single survey which was sent to participants via email.

A subsample of 275 participants took part in Study 2 (66% female, 34% male), with a mean age of 26 years (SD = 10.01), and 15.02 years of formal education (SD = 2.96). We evaluated participants’ social cognition and executive functioning abilities. We carried out this evaluation through a videoconference. We invited all participants from study 1 to take part in study 2, hence participants in the second sample were selected due to convenience.

For Study 1, all participants gave their informed consent in accordance with the Declaration of Helsinki by pressing an “I agree” button located beneath an explanatory letter. For study 2, participants consented once again before proceeding with the evaluation. Institutional Ethics Committee from the Universidad Antonio Nariño revised and approved the research project (approval # 20,190,516). We collected the data from both studies between July 2021 and June 2022.

Instruments

Study 1

Data for this study was collected between October 2020 and June 2022. Participants first reported their age, sex, occupation, and city of residence and then completed the scales, as described below.

The Hospital Anxiety and Depression Scale (HADS) was used to screen the presence of clinical levels of anxious symptomatology (Zigmond & Snaith, 1983). This scale has 14 questions, divided into two sub-scales, anxiety and depression. Questions were answered according to a 4-point Likert scale. The present study used scores of eight and above for the anxiety subscale as the cutoff point indicating clinical levels of anxious symptomatology. This cutoff point has been frequently used in previous research (Brennan et al., 2010). The version of the scale we used in our study had been previously translated and validated for its use in Spanish, with a good internal consistency of of α = 0.85 (Herrero et al., 2003).

The Zung Self-rating Anxiety Scale (ZSAS) measured clinical levels of anxiety, based on the symptoms listed in the clinical criteria for this disorder (Zung, 1971). This scale has 20 questions, answered through a 4-point Likert scale. Clinical levels of anxiety were indicated by scores of 36 or above (Dunstan & Scott, 2020). It has been previously validated and used with Spanish-speaking participants, with good internal consistency by Cronbach alpha α = 0.83 (De La Ossa et al., 2009).

The Short-Form Health Survey 12 (SF-12) was used to measure self-perception of quality of life regarding physical and mental health (Ware et al., 1996). This scale has eight subscales: physical functioning, physical role, bodily pain, general health, vitality, social functioning, emotional role and mental health. It consists of 12 questions, with diverse response types (3, 5 and 6-point Likert scales, and yes or no questions). Indices for physical and mental health were calculated by adding the obtained scores, such that higher scores indicate better health (Hagell et al., 2017). We used a translated and validated version of the SF-12 (Monteagudo Piqueras et al., 2009), with an internal consistency of α = 0.82 (Ramírez-Vélez et al., 2010).

The List of Threatening Events (LTE) is a questionnaire used to evaluate and quantify stressful events that have occurred throughout the participant’s life (Brugha & Cragg, 1990). This self-report instrument consists of 12 questions, each stating a stressful life event that the person could have experienced, accompanied by yes/no response alternatives. The total score of the scale was obtained by adding all items to which response was yes, indicating the number of stressful life events experienced. The LTE has been validated and used with Spanish-speaking participants, with an internal consistency of α = 0.44 (Motrico et al., 2013).

Study 2

Measures for the second study were collected between July 2021 and June 2022. Participants from Study 1 who agreed to participating in Study 2, were scheduled for a video conference via Zoom, and administered the following tests:

Executive functions

We used the digit spans subtest, from the Weschler Adult Intelligence Scale (WAIS), to evaluate working memory (Hill et al., 2010). We used the raw score from the inverse digits task to measure this domain. The digit spans subtest has a good internal consistency of α =.82 (Palacio et al., 2022; Henao-Arboleda et al., 2010). The verbal fluency task measured the participants’ capacity to initiate and sustain word production. The specific task we used for this study was a phonological verbal fluency task using the letters M and P, since it has most frequently been used when working with Spanish-speaking participants (Peña-Casanova et al., 2009), with good test-retest reliability with ICCs of 0.77 (Martino et al., 2022). The final score corresponded to the number of unique and correct words achieved.

Social cognition

We used a short version of the Emotion Evaluation Test (EET). The EET is the first part of The Awareness of Social Inference Test (TASIT) and measured the participant’s capacity to recognize others’ emotions (McDonald et al., 2003). We presented 10 video clips in which the actors represented anger, disgust, fear, sadness and surprise in everyday social interactions. At the end of each clip, the participant had to select the emotion that was demonstrated. We obtained the final score by adding the number of clips in which the participant correctly identified the emotion. The materials we used for this test had been translated and adapted to Spanish (Baez et al., 2014), with an internal consistency of α = 0.69 (Palacio et al., 2022).

The Reading the Mind in the Eyes Test (RMET) was used to evaluate the ability to identify others’ mental states (Baron-Cohen et al., 1997). We presented participants with 25 different photographs of people, in which only the eyes region was shown. We asked participants to select the word, among for options, which could best describe what the person in the image was feeling or thinking. To obtain the final score we added all correct answers. This test has been previously translated and validated for its use in Spanish (Roman et al., 2012), and has a good internal consistency of α = 0.853 (Palacio et al., 2022).

Data analysis

Study 1

We calculated a Global Anxiety variable by adding the scores obtained in the anxiety subscale of the HADS and the ZSAS, to obtain a single definitive measure of each participant’s anxiety level. Since both scales are interpreted in the same direction, this approach allowed us having a more robust measure for anxiety, as observed in previous studies (Çiçekci et al., 2023). To explore the relationship between anxiety levels and demographic aspects, health and threatening life events variables, we conducted a linear regression model. In this model, we considered global anxiety scores as a dependent variable, and demographic aspects (sex, age and education years), health indices (SF-12 physical and mental health indices) and number of threatening life events (LTE score) as predictors.

In addition, we explored the differences in health indices and number of threatening life events between participants with different levels of anxiety. Thus, participants were divided into three groups according to the scores obtained in both anxiety scales [low anxiety symptoms (HADS: 0–6, ZSAS: 20–35), mild symptoms (HADS: 7–10, ZSAS: 36–45), and moderate-severe (HADS: ≥ 11, ZSAS: ≥ 46). Then, we calculated one-way ANOVAs to compare the SF-12 and LTE scores between participants who had moderate-severe, mild and no anxiety symptoms.

Study 2

For this study, we calculated the executive functions score by adding the scores obtained in the verbal fluency and inverse digits tasks. We also calculated a social cognition accuracy score by averaging participants’ percentage of correct responses in the TASIT and the RMET.

To assess the relationship between anxiety levels and executive functions and social cognition, we performed a linear regression model using global anxiety as a dependent variable and composite scores for social cognition and executive functions as predictors. To explore whether results obtained in the Study 1 replicated in Study 2 we ran a second model maintaining these variables as predictors, but also including demographics, health indices (SF-12) and the number of stressful life events experienced (LTE). Finally, we calculated ANOVA’s between the same participant groups as in Study 1, to compare the social cognition and executive functions performances between groups with different levels of anxiety.

For all analyses, we considered a significance level of p < 0.05 and used the software JASP 0.16.0.0. To calculate effect size, we used partial \({{\upeta }}^{2}\) and interpreted effect sizes of 0.0099 as small, 0.0588 as medium and 0.1379 as large (Richardson, 2011).

Results

Study 1

Participants having scores above the cutoff points for the anxiety scales in Study 1 were 302 for the HADS (55%; M = 8.5; DS = 4.5), 353 for the ZSAS (64.4%; M = 40; DS = 9.9), and 276 for both scales (50.4%).

The linear regresion model with global anxiety as dependent variable was significant (\({r}^{2}\)= 569, f [4,540] = 120.922, p < 0.05) and predicts 56.9% of the variance in participants anxiety scores. Sex, age and physical and mental health indices significantly predicted anxiety symptoms. The number of stressful life events experienced, and the years of formal education were not significant predictors (Table 1). Both physical and mental health indices were the strongest predictors for participants anxiety levels (Fig. 1A and B).

Table 1 Coefficients for Study 1 global anxiety regression model
Full size table
Fig. 1
figure 1

Linear regression model and group comparison results for Study 1. A Linear regression model with physical health index as a predictor for anxiety symptoms, B Linear regression model with mental health index as predictor for anxiety symptoms, C Mean differences in SF-12 scores between participants with none, mild and moderate- severe symptoms in both anxiety scales, *** = p < 0.001. D Mean differences in LTE scores between participants with none, mild and moderate- severe symptoms in anxiety scales, *** = p < 0.001

Full size image

Both physical and mental health indices differ significantly between participant groups, with large effect sizes in both cases (F [2,358] = 111.683, p < 0.001, partial \({{\upeta }}^{2}\)= 0.384; F [2,358] = 167.121, p < 0.001, partial \({{\upeta }}^{2}\)= 0.483) (Fig. 1C). Participants with no symptoms had significantly higher physical and mental health scores than participants with mild and moderate-severe symptoms (t [1] = 7.517, p < 0.001; t [1] = 10.299, p < 0.001; t [1] = 14.874, p < 0.001; t [1] = 18.022, p < 0.001) and those with mild symptoms also had significantly higher scores than those in the moderate-severe group (t [1] = 6.148, p < 0.001; t [1] = 6.313, p < 0.001).

Participants with different levels of anxiety also differed significantly in terms of the number of threatening life events experienced, with a small effect size (F [2,358] = 6.878, p < 0.05, partial \({{\upeta }}^{2}\)= 0.037) (Fig. 1D). The Tukey post-hoc test indicated that participants with no symptoms had significantly lower LTE scores than participants with moderate-severe symptoms (t [1] =-3.624, p < 0.001), while there were no significant differences between participants with mild symptoms and those with no and moderate-severe symptoms ( t [1] = -2.226, p = 0.068); (t [1] =-1.122, p = 0.501).

In sum, age, sex and health indices significantly predict participants’ anxiety scores, with health indices being the strongest predictors, while stressful life events were not significant in this regard. Both health indices and events experienced are associated with the severity of anxiety symptoms, although the effect of health indices is more robust.

Study 2

Participants having scores above the cutoff points for the anxiety scales in Study 2 were 158 for the HADS (57.5%; M = 8.8; DS = 4.5), 184 for the ZSAS (66.9%; M = 41; DS = 10), and 145 for both scales (52.7%).

The linear regression model including exclusively executive functions and social cognition as predictors was significant (\({r}^{2}\)= 0.025, f [2,272] = 4,479, p < 0.05), although it predicted only 2.5% of the variance, with executive functions as the only significant predictor (β= -0.131, t[274] =-2.192, p < 0.05) (Fig. 2A).

Fig. 2
figure 2

Linear regression models results for Study 2. A Linear regression model with executive functions scores as a predictors for anxiety symptoms, B Linear regression model with physical health index as predictor for anxiety symptoms, C Linear regression model with mental health index as predictors for anxiety symptoms, D Linear regression model with LTE scores as predictors for anxiety symptoms, E Linear regression model with social cognition scores as predictors for anxiety symptoms

Full size image

The second model, which also included predictors used in the Study 1 was also significant (\({r}^{2}\)= 0.584, F [8,266] = 49.175, p < 0.05). When comparing both models we found that the second model is a better fit since it predicts 58.4% of the variance of participants’ anxiety scores. This model explains less than 2% more variance than the one presented in Study 1, indicating that variables from study 2 explained only a small part of the variance. The same predictors as in the two previous models were significant (age, sex, health indices and executive functions scores) (Fig. 2B, C, D and E), but in addition the number of stressful life events (LTE) and social cognition scores were also significant (Table 2). As in Study 1, the health indices (SF-12) were the strongest predictors of anxiety.

Table 2 Coefficients for Study 2 global anxiety regression model including all predictors
Full size table

Executive functions scores differ significantly between participant with different anxiety levels, with a small effect size (F [2,174] = 3.438, p < 0.05, partial \({{\upeta }}^{2}\)= 0.038) (Fig. 3A). Participants with no symptoms had significantly higher executive functions scores than participants with moderate-severe symptoms (t [1] =-2.621, p < 0.05), and we did not find significant differences between participants with mild symptoms and those with no and moderate-severe symptoms (t [1] = 1.268, p = 0.415); (t [1] = 1.154, p = 0.482). Social cognition scores did not differ significantly between participant’s groups (fF[2,174] = 1.917, p = 0.150) (Fig. 3B).

Fig. 3
figure 3

Group comparisons for Study 2. A Mean differences in executive functions scores between participants with none, mild and moderate- severe symptoms in anxiety scales, *= p < 0.05. B Mean differences in social cognition scores between participants with none, mild and moderate- severe symptoms in anxiety scales

Full size image

Summarizing, health indices, age and sex predicted more than 50% of participants’ anxiety scores, and this finding was corroborated through both studies. The stressful life events, executive functions and social cognition also predict these scores in study 2, although the percentage of variance explained is considerably smaller. Participants experiencing no anxiety symptoms presented higher executive functions scores than those who had mild and moderate-severe symptoms, while social cognition performance did not vary regarding symptom severity.

Discussion

Our research aimed to investigate the relationship between various anxiety-related variables and the severity of symptoms in adults. In Study 1, we examined the connection between these symptoms and health indices, as well as stressful life events. In Study 2, our focus shifted to executive functions and social cognition. Our findings demonstrated that physical and mental health emerged as the most influential factors in predicting the severity of anxiety symptoms, effectively distinguishing between different levels of anxiety. In contrast, the associations between anxiety symptoms and variables such as stressful life events, executive functions, and social cognition were relatively weaker. These valuable insights contribute to our understanding of the significance of these variables when it comes to anxiety and could be useful when exploring prevention and treatment of subthreshold anxiety.

We found a high prevalence of anxiety symptoms within the sample, with at least half of the participants passing the cutoff points for both scales. Reviews focusing on the prevalence of anxiety disorders and subthreshold anxiety have indicated estimates of 10.6% and between 4.2 − 12.4% respectively (Haller et al., 2014; Remes et al., 2016). Research focusing on anxiety prevalence at time periods surrounding the COVID-19 pandemic have estimated an incidence between 31 and 35% (Chang et al., 2021; Deng et al., 2021; Zhang et al., 2022). Since measures for anxiety symptoms for the current study were taken after the quarantines, we may assume that the elevated measures of anxiety symptoms we found were partially influenced by this context.

Results of Study 1 showed that mental and physical health indices were strong predictors of anxiety symptoms, with participants with more severe anxiety presenting poorer health outcomes than those with mild anxiety and no anxiety. Consistent with these findings, previous studies have shown that mental health status and psychological wellbeing are significantly associated with the presence and severity of anxiety symptoms (Faisal et al., 2022; Rapheal & K, 2014), with stronger associations found between GAD and PD and the mental health-related quality of life (Comer et al., 2011). Likewise, associations between poorer physical health and anxiety are coherent with findings of previous studies. Studies focusing on anxiety and physical health have explored and found associations between both objective health, i.e., chronic and specific medical conditions (Himmelfarb & Murrell, 1984; Jones et al., 2017), and subjective health, understood as the perception a person has regarding their health (Liao et al., 2021; Strine et al., 2005). However, associations between subjective health and anxiety appear to be more consistent (Mehta et al., 2007; Paukert et al., 2010). Since the measures for physical health indices, we used in the current study were subjective, our findings support the hypothesis that perceived impairment in health or poorer health related quality of life could be relevant to understanding the occurrence of anxiety symptoms. Due to the measures selected, no conclusions are drawn regarding objective physical health in relation to anxiety, although previous research would indicate that it is directly implicated as well (Tian et al., 2023).

Results regarding the relationship between anxiety and stressful life events were mixed. In Study 1, the numbers of stressful events did not predict anxiety, while in Study 2 we found a significant but weak association. In addition, anxiety severity groups did differ significantly in terms of the number of stressful events. Our findings suggest that there is some positive association between stressful life events and severity of anxiety symptoms (Schneider et al., 2021; Tiguman et al., 2022). The nature of this association could be additive, where a higher number of events experienced resulted in more severe symptoms as seen in the differences between groups. This is in coherence with previous findings which have suggested that stressful events and vulnerabilities add to each other and increase the probability of anxiety or other mental health conditions (Beurs et al., 2001). The type of events experienced could also result in distinct symptom and symptom severities, for example differences between childhood trauma and events in later life. Events of physical, psychological and sexual abuse have been found as early events resulting in future anxious arousal and distress, (van Veen et al., 2013); whereas serious financial problems, serious conflict with significant others and unemployment appear to be main triggers for anxiety in adulthood (Beurs et al., 2001; van Veen et al., 2013). Hence, it is likely that certain types of events, and their timing throughout the lifespan, result as stronger predictors of anxious symptoms. We relied on the LTE to measure stressful events in adult life exclusively, hence employing multiple instruments to cover a wider range of events across the lifespan could be useful to explore this differential impact in future studies.

Regarding executive functions, we found that they were a weak predictor for anxiety symptoms. However, they do serve to differentiate groups based on severity, with executive dysfunction being more pronounced in individuals with moderate-severe anxiety, when compared to those with no anxiety (Parra-Díaz et al., 2021; Sharp et al., 2015). It is possible that, since adequate executive functioning is required to self-regulate behavior and cognition, then an impairment in these functions may result in greater difficulties to cope with anxiety generating situations, for example when inhibiting anxiety reinforcing responses or attempting to shift attention away from anxious stimuli (Sharp et al., 2015). On the other hand, the presence of elevated anxiety symptoms may difficult executive functions performance, as anxiety can operate as a distractor which occupies part of a persons’ attentional capacities or biases attention away from goal-oriented activities (Majeed et al., 2023; Palomares Castillo et al., 2010). Further research would be useful to establish the direction of the link between executive dysfunction and anxiety symptoms. In the current study, we measured executive functions regarding working memory and verbal fluency. Both of these processes have been found, although not consistently, to be affected in individuals with anxiety (Ajilchi & Nejati, 2017; Gulpers, 2019; Nyberg et al., 2021; Toazza et al., 2016). Previous studies suggest that anxiety can affect performance in tasks associated with these processes, mainly due to time and success pressures. Still, an impairment in day-to-day tasks involving working memory and verbal fluency could result in elevated anxiety symptoms, as it could generate difficulties navigating social and work-related settings. Other processes within executive functions have been found to be associated with anxiety, such as planning and inhibitory control (Fonoff et al., 2015; Predescu et al., 2020), hence, further studies could include them as part of an executive functions overall evaluation.

Social cognition showed a weak association with anxiety symptoms, implying that individuals with lower emotion recognition and theory of mind abilities are more likely to present subthreshold anxiety. Previous findings regarding the association between these variables have been mixed, which is not surprising given the association found was the weakest one among the studied variables (Hollocks et al., 2014; Mavrogiorgou et al., 2016; Nikolić, 2020). In addition, social cognition did not appear to be a variable useful to differentiate between anxiety severity groups. This discrepancy could indicate that social cognition presents some general association to anxiety, but that the measures used were not sensitive enough to differentiate between distinct anxiety severity groups. The associations we found indicate that the social cognition domains we contemplated in our study, mainly the ability to recognize others’ emotions and mental states, could be relevant to understanding certain impairments associated with anxiety symptoms. These difficulties in recognition have been found to result in impairments in navigating daily social interactions, actual or perceived diminished social competence and difficulties in social integration (Baez et al., 2023; Plana et al., 2013). Hence, contexts involving social interaction, such as work, school, family, dating, among others, could become anxiety generating situations for people presenting these impairments, increasing the possibility of presenting elevated anxiety symptoms and behaviors, such as worrying, ruminating and experiential avoidance.

Sex and age were also significantly related to anxiety symptoms within our sample, with women and younger participants presenting higher anxiety scores. Previous findings focusing on demographic variables have for the most part indicated similar results. Women can be almost twice as likely to present several anxiety disorders as well as a higher prevalence of subthreshold GAD (Gustavson et al., 2018; Haller et al., 2014; Remes et al., 2016). Similar to our sample anxiety prevalence tends to decrease with age, with people in their twenties presenting higher levels of anxiety than those in their late thirties and forties (Gustavson et al., 2018; Remes et al., 2016; Scott et al., 2008).

The findings are consistent with the study’s sample, which comprised the general adult population of Colombia. The Colombian National Health Care Survey has shown a nationwide increase in symptoms of anxiety (Gómez-Restrepo et al., 2016; Martínez-Cabezas et al., 2023), with a more pronounced escalation observed during and following the COVID-19 pandemic (Lapeira-Panneflex et al., 2023; Uribe-Restrepo et al., 2022, 2022). Specific psychosocial stressors in the Colombian context, such as internal armed conflict, forced displacement, poverty, and violence, are significant stressful life events that have been linked to heightened anxiety levels across various groups (Giebel et al., 2023; Mejía et al., 2024; Richards et al., 2011). Furthermore, barriers to accessing mental health services in Colombia have been identified, both systematic, in terms of service implementation and availability across socioeconomic strata, and attitudinal, related to prejudices encountered when seeking these services (Agudelo-Hernández & Rojas-Andrade, 2023; Campo-Arias et al., 2020). These obstacles have hindered the effective prevention and treatment of anxiety disorders, particularly among lower-income and rural communities.

It is worth noting that there are several factors which can relate to anxiety which go beyond the scope of this study, including other clinical symptoms and conditions, personality traits, quality of life, among others (Antony et al., 1998; Gorman, 1996; Olatunji et al., 2007). Future studies could encompass a wider variety of these factors and examine how they relate to each other and how they contribute to predicting and preventing elevated levels of anxiety.

Limitations

Our study had considerably more women than men (66% of women to 34% of men), and despite having an age range of 18–73 years, the average age of our participants was only of 26 years. While our study boasts a broad age range of 18–73 years, the average age of participants was notably lower at 26 years. This discrepancy is largely attributable to a selection bias, since we implemented convenience sampling to invite participants to take part in our study. This bias resulted in a high proportion of university students within our participant pool, which skews the average age downwards. This imbalance in proportion of participants in terms of sex and age was not intended but could have an impact over the findings obtained for each of the variables studied. Hence, balancing these groups or investigating the differential impact of sex and age with respect to anxiety symptoms could be important, since it might yield different results. Future studies should aim to recruit a more representative sample to ensure findings are reflective of broader age demographics.

We identified other limitations to our study, mainly affecting study 2. Firstly, the measures included for executive functions and for social cognition may have been limited in evaluating these constructs. As for executive functions, only measures of working memory and verbal fluency were included, hence processes such as attention, inhibition and planning were not considered, resulting in an inaccurate measurement of the construct. Similarly, social cognition measures included were recognizing emotion and mental states in others, which is important but not the only components within social cognition. In addition, the measures used for this domain could also be influenced by recall bias, since these tasks may allow participants to provide answers based of previous responses, so the use of complementary measures which are not subject to this bias would allow a more precise evaluation of social cognition.

Finally, the sample size was greatly reduced from study 1 to study 2, due to participants’ availability to continue in the study, which impacts statistical analysis regarding the detection of small effect sizes of the mentioned variables. This reduction in the sample may also result in an attrition bias, since participants who remained for the study 2 measures may have common characteristics influencing the results from the study.

Implications and future directions

In sum, our study allowed us to clarify the relevancy of the different variables of interest for the study of anxiety in the general adult population. A decline in health or perceived health could be relevant when preventing the development of anxiety symptoms, whereas the experience of stressful life events and impairments in executive functions are mildly relevant for this purpose but still worth considering. The former hints at the importance that each of these variables could have within anxiety prevention programs and in the development of psychoeducational material regarding anxiety symptoms, as well as in psychological treatment. Prioritizing health variables within these processes appears to be the most relevant action when attempting to better approach anxiety (Mehta et al., 2007).

It will be important to carry out studies with larger samples with a similar focus to the present one, to be able to study more rigorously the relationship between anxiety symptoms and relevant variables such as health indices, stressful life events and executive functioning. Specifically, regarding health indices, future research could attempt to differentiate the relationship between anxiety symptoms and objective and subjective measures of both physical and mental health. The current study focused mainly on subjective measures, but having both into account would allow a better understanding and intervention approaching this variable.

Studying the impact of stressful life events on anxiety symptoms using instruments which cover a wider range of events across the lifespan is another direction for future research. The use of a comprehensive set of assessment tools that encompass events experienced from childhood through adulthood, would provide a more comprehensive understanding of how different types and timing of stressful or traumatic events may contribute to the development and severity of anxiety symptoms (van Veen et al., 2013).

Finally, since both executive functioning and social cognition are both psychological constructs composed of diverse processes, future research could focus on exploring the relationship between anxiety and these specific processes, to distinguish which of them are relevant for its prevention and intervention. In addition, since both variables have shown to be more strongly related to anxiety in populations different to the general adult population (Fonoff et al., 2015; Hollocks et al., 2014; Pearcey et al., 2021), studies comparing between populations would be informative.

Conclusion

Variables mildly studied previously in relation to anxiety are relevant to its understanding and could also be useful for its treatment and prevention. Among the variables we studied here, both studies indicated that health indices are especially important to this aim, with stressful events, executive functioning and social cognition still requiring further research to identify their specific relevance.