Depression, Anxiety, and Stress in a Primary Care Population in Northern Portugal: A Cross-Sectional Study in the Post-COVID-19 Context

Abstract

Background: The COVID-19 pandemic has had lasting psychological consequences, with primary care playing a key role in identifying emotional distress. Objective: Our objective was to characterize levels of depression, anxiety, and stress in a primary care population in the post-pandemic context and to examine their associations with sociodemographic and COVID-19-related variables. Methods: A cross-sectional study was conducted among 260 adult users of a Family Health Unit. Participants completed a sociodemographic questionnaire and the Depression, Anxiety, and Stress Scale (DASS-21). Group differences were assessed using Mann–Whitney U and Kruskal–Wallis tests, and multivariable generalized linear models (negative binomial regression for symptom scores and logistic regression for perceived impact) were used to identify factors independently associated with each outcome. Results: Most participants fell within normal symptom ranges (66.9% depression; 65.0% anxiety; 74.6% stress), although approximately one-third presented at least moderate levels of depressive or anxiety symptoms. In unadjusted analyses, women, older adults and professionally inactive individuals had significantly higher levels of psychological distress. Self-reported COVID-19 infection history was not associated with DASS-21 scores. In contrast, participants reporting a negative psychological impact from the pandemic and poorer self-rated mental health had significantly higher depression, anxiety, and stress scores. Conclusions: In this primary care sample, self-reported psychological impact from the pandemic and self-rated mental health were more strongly associated with psychological distress than self-reported COVID-19 infection history. Multivariable analyses identified independent associations between psychotropic medication use, self-rated health, and DASS-21 scores and the outcomes examined. These findings indicate associations between these variables within the cross-sectional sample; however, their temporal or causal relationships cannot be established.

1. Introduction

The Coronavirus Disease 2019 (COVID-19) pandemic represented an unprecedented global public health crisis, characterized by widespread mortality, stringent public health measures, such as lockdowns and social isolation, and profound socioeconomic disruption [1]. Beyond the immediate physical health consequences, the pandemic and its aftermath have been consistently associated with a substantial and global increase in adverse mental health outcomes [2,3].

Early in the pandemic, systematic reviews consistently reported a substantial increase in symptoms of depression, anxiety, and stress in the general population. For example, Xiong et al. [4] reviewed 19 studies and found prevalence estimates ranging from 14.6% to 48.3% for depression, 6.3% to 50.9% for anxiety, and 8.1% to 81.9% for stress, reflecting considerable heterogeneity across countries, study populations, and assessment instruments. This variability highlights the influence of contextual factors and methodological differences, making direct comparisons between studies challenging.

More recent evidence suggests that psychological distress has persisted beyond the acute phase of the pandemic, although prevalence appears to have declined compared with the earliest stages. A recent systematic review of European primary care populations reported that clinically relevant symptoms of depression, anxiety, and stress remain common in the post-pandemic period, but estimates varied substantially because of differences in study design, participant characteristics, and outcome measures [2]. Likewise, recent reviews have emphasized that psychological recovery has been heterogeneous, with persistent vulnerability observed among specific population groups rather than uniformly across the population [3,5].

Although numerous studies have documented elevated levels of depression, anxiety, and stress during the acute phases of the COVID-19 pandemic, several important gaps remain. First, much of the available evidence was generated during the early stages of the pandemic, when restrictive public health measures and uncertainty were at their peak, limiting its applicability to the current post-pandemic context. Second, many studies have focused on the general population or specific occupational groups, such as healthcare workers, while comparatively little attention has been paid to primary care users, despite this setting being central to the identification and management of common mental health problems. Third, previous studies have produced inconsistent findings regarding the relationship between SARS-CoV-2 infection and subsequent psychological distress, suggesting that subjective experiences of the pandemic may be more relevant than infection history alone. These limitations highlight the need for studies examining psychological well-being in primary care populations during the post-pandemic period.

Therefore, this study aimed to characterize the symptoms of depression, anxiety, and stress among adult users of a Family Health Unit in Northern Portugal during the post-pandemic period and to examine their associations with selected sociodemographic, clinical and COVID-19-related factors identified in the previous literature. Based on the previous literature, we hypothesized that: (1) symptoms of depression, anxiety, and stress would remain common in the post-pandemic period; (2) women, professionally inactive participants, and individuals with a previous psychiatric disorder or psychotropic medication use would report higher symptom levels; (3) participants reporting a negative psychological impact of the pandemic and poorer self-rated mental health would have higher DASS-21 scores; and (4) self-reported COVID-19 infection history would show a weaker association with psychological symptoms than participants’ subjective experience of the pandemic.

2. Materials and Methods

2.1. Study Design, Duration, and Setting

A descriptive, observational, analytical cross-sectional study was conducted at a single Family Health Unit located in Northern Portugal (Unidade de Saúde Familiar Fénix—USF). Participants were recruited consecutively between 1 February 2025 and 31 January 2026, among adult users attending the participating primary care setting during the study period. A total of 260 individuals were approached for participation, and all met the predefined eligibility criteria and agreed to participate. Consequently, all 260 participants were included in the final analysis, with no exclusions or non-participation. Although ethical approval was granted in April 2025, participant recruitment commenced in 2026. As this was an investigator-initiated, non-funded study conducted alongside the researchers’ routine academic and clinical responsibilities, additional time was required to complete the logistical preparation and coordinate data collection with the participating Family Health Unit.

This cross-sectional observational study was designed and reported in accordance with the Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) Statement [6]. A completed STROBE checklist is provided as Supplementary Material.

2.2. Study Population

The study population comprised all users registered at the Unidade de Saúde Familiar Fénix (n = 13,550) who attended the unit during the study period and met the established inclusion and exclusion criteria.

2.3. Sample Size

The minimum required sample size was calculated using a proportional formula for finite populations, assuming a 95% confidence level and a margin of error of 5.86%. This resulted in a minimum sample size of 250 participants.

2.4. Eligibility Criteria

2.4.1. Inclusion Criteria

Participants were eligible for inclusion if they:

• Were registered users of the USF;
• Were aged 18 years or older;
• Attended the health unit on the scheduled days for data collection;
• Provided written informed consent to participate in the study.

2.4.2. Exclusion Criteria

Users were excluded if they demonstrated cognitive impairment or difficulty understanding concepts that prevented adequate comprehension of the questionnaire or the expression of informed opinions.

2.5. Data Collection Procedures

Data were collected using a standardized questionnaire consisting of a socio-demographic section and validated psychometric scales. The questionnaire was primarily administered digitally via Google Docs^®, with supervised access provided by a study investigator. Subsequently, the collected data were organized in a Microsoft Excel^® (Microsoft Office 2016) database.

For participants with limited familiarity with digital technologies or those who requested assistance, alternative data collection methods were provided, including paper-based questionnaires or in-person assistance from a trained investigator, ensuring equal access and minimizing selection bias.

2.6. Study Variables

Based on the predefined research hypotheses, the following variables were collected and operationalized:

• Age: discrete quantitative variable, expressed in years;
• Gender: nominal variable (male/female/other);
• Marital status: nominal variable (single, married/cohabiting, divorced, widowed);
• Educational level: ordinal variable, categorized as low (no formal education or up to 9th grade), medium (9th to 12th grade), or high (university degree or equivalent);
• Employment status: nominal variable categorized as professionally active (regular paid employment) or professionally inactive (unemployed, retired, or student). These categories were combined because of the small number of participants in some individual employment categories, which would have limited the statistical power and stability of subgroup analyses. This categorization was adopted solely for analytical purposes and is acknowledged as a limitation of the study;
• Self-reported COVID-19 infection history: nominal variable based on participants’ self-report (reported never having had COVID-19; reported previous COVID-19 infection without hospitalization; reported previous COVID-19 infection with hospitalization). No laboratory confirmation or review of medical records was performed;
• Perceived impact on mental health: nominal variable (yes/no);
• Self-rated mental health after the COVID-19 pandemic: ordinal variable (very good, good, fair, poor);
• Self-rated overall health quality: ordinal variable (very poor, fair, good, very good);
• Psychotropic medication use: nominal variable (yes/no), considered affirmative if medication was initiated;
• Pre-existing medical condition: nominal variable (yes/no), considered affirmative if a prior diagnosis of a mental disorder (anxiety, depression, schizophrenia, bipolar disorder, or psychosis) was reported.

2.7. Data Collection Instruments

2.7.1. Depression, Anxiety, and Stress Scale (DASS-21)

The DASS-21 Portuguese validated version was used to assess the severity of symptoms of depression, anxiety, and stress. The instrument comprises three subscales—Depression, Anxiety, and Stress—each containing seven items. Participants rate the extent to which each statement applied to them over the previous week using a 4-point Likert scale ranging from 0 (“Did not apply to me at all”) to 3 (“Applied to me very much or most of the time”).

Scores for each subscale were calculated by summing the responses to the corresponding seven items and multiplying the total by two, following the standard DASS-21 scoring procedure, yielding scores comparable to the original 42-item DASS. Higher scores indicate greater symptom severity. For descriptive analyses, scores were categorized into normal, mild, moderate, severe, and extremely severe according to the validated Portuguese cut-off values.

The DASS-21 demonstrated excellent internal consistency in the present sample. Cronbach’s alpha coefficients were α = 0.935 for the Depression subscale, α = 0.907 for the Anxiety subscale, α = 0.940 for the Stress subscale, and α = 0.971 for the total scale.

2.7.2. Perceived Psychological Impact of the Pandemic

The perceived psychological impact of the pandemic was assessed using the question: “Do you believe that the COVID-19 pandemic had a negative impact on your mental health?” (yes/no). This item referred to participants’ perception of the pandemic’s impact at any point during the pandemic.

2.7.3. Self-Rated Mental Health After the Pandemic

The self-rated mental health after the pandemic was assessed using the question: “How would you rate your current mental health after the COVID-19 pandemic?” Response options were very good, good, fair, and poor. This item referred to participants’ current mental health status at the time of questionnaire completion.

2.8. Application of the Instruments

Questionnaires were administered in a private consultation room by trained investigators who were clearly identified to participants. After confirming eligibility, the study objectives were explained, and written informed consent was obtained. Participants were informed that all data collected was anonymous and confidential and used exclusively for research purposes.

Participation was entirely voluntary. Potential participants were informed that their decision to participate or decline would not affect the healthcare they received at the Family Health Unit in any way. Recruitment was conducted only after eligibility had been confirmed, and questionnaires were completed in a private setting to ensure confidentiality and minimize any perceived pressure to participate. Participants were also informed that they could withdraw from the study at any time without providing a reason.

2.9. Statistical Analysis

Data analysis was performed using the Statistical Package for the Social Sciences (SPSS^®), version 29.0 (IBM Corp., Armonk, NY, USA) and R (version 4.5.1). Independent variables were treated as categorical. The variables included in the analyses were selected a priori based on previous evidence suggesting their potential association with symptoms of depression, anxiety, and stress in primary care and post-pandemic populations.

Scores from the DASS-21 subscales (depression, anxiety, and stress) were analyzed as continuous variables for mean comparisons and, additionally, categorized into severity levels (normal, mild, moderate, severe, and extremely severe) for descriptive purposes, according to the validated Portuguese cut-off points.

The normality of the DASS-21 score distributions was assessed using the Kolmogorov–Smirnov and Shapiro–Wilk tests. Because the normality assumption was violated, the response variables were markedly skewed, and several subgroups of the explanatory variables comprised a small number of participants, non-parametric tests were used for all group comparisons.

Differences between two independent groups were assessed using the Mann–Whitney U test, while differences among three or more groups were assessed using the Kruskal–Wallis test and H statistic value.

Effect sizes were quantified: absolute rank-biserial correlation, r, for the Mann–Whitney test and epsilon squared (ε²) for the Kruskal–Wallis test. Statistical significance was set at a two-tailed p-value < 0.05.

Post hoc power analyses were conducted only for comparisons that reached statistical significance. Statistical Power (1 − β_II) was estimated with G*Power (version 3.1.9.7; Heinrich Heine University Düsseldorf, Düsseldorf, Germany) on the basis of the observed sample size and effect size. Because the inferential analyses were performed with non-parametric tests, owing to violation of the normality assumption, the power calculations were carried out within the corresponding non-parametric framework, using approximation of the observed effect size d to the Mann–Whitney test and Cohen’s f to Kruskal–Wallis test results.

2.9.1. Multivariable Analysis

Associations between mental health outcomes and sociodemographic, COVID-19-related, and health-related characteristics were evaluated using generalized linear models (GLMs) fitted in R. The sociodemographic predictors were age group (<45 years as reference), gender (female as reference), employment status (professionally inactive as reference), and educational level (up to the 9th grade as reference). Self-reported COVID-19 infection history, psychotropic medication use after COVID-19 (no as reference), and self-rated health after COVID-19 (poor as reference) were also considered as potential explanatory variables. For modelling purposes, the educational categories with the lowest frequencies were collapsed into a single category corresponding to education up to the 9th grade. This recategorisation was performed to improve model convergence and to ensure more stable parameter estimates, given the small number of observations in some of the original educational categories.

Due to the asymmetric and count-like nature of the depression, anxiety, and stress scores, several plausible generalized linear model specifications were evaluated. Negative binomial regression with a log link and Gamma regression were considered, with log and identity link functions evaluated for the latter. Gamma models were fitted after adding one unit to the scores to accommodate zero values. Models were also evaluated under progressively adjusted sets of covariates, including sociodemographic characteristics alone, followed by the inclusion of factors such as self-reported history of COVID-19 infection, psychotropic medication use, and self-rated mental health after COVID-19 infection. Perceived impact on mental health was analysed as a binary outcome using multivariable logistic regression with a logit link function.

The significance of individual model parameters was assessed using Wald tests. For categorical predictors with more than two levels, significant overall effects were followed, where appropriate, by Tukey-adjusted pairwise comparisons to identify differences between factor levels.

For each candidate model, model adequacy was assessed using information criteria and model-fit and prediction measures, including Akaike’s Information Criterion (AIC), residual deviance, root mean squared error (RMSE), and the correlation between the observed and fitted values. Bidirectional stepwise selection was then applied using the Bayesian Information Criterion (BIC) to obtain parsimonious models from the prespecified candidate variables. Alternative model specifications were examined to assess the robustness of the estimated associations with respect to the assumed distribution, link function, and degree of covariate adjustment. The final model specification was selected based on an overall assessment of model adequacy and consistency of the estimated associations across alternative specifications.

For the negative binomial models, regression coefficients (β) were exponentiated and are presented as exp(β), together with their 95% confidence intervals. Values of exp(β) represent the multiplicative change in the expected depression, anxiety, or stress score associated with each predictor, conditional on the other variables included in the model. Thus, values greater than 1 indicate higher expected scores, whereas values below 1 indicate lower expected scores. For the logistic regression models, exponentiated coefficients were interpreted as adjusted odds ratios (aORs) and are presented with 95% confidence intervals.

Model performance for the negative binomial models was evaluated using the Akaike Information Criterion (AIC), residual deviance, root mean squared error (RMSE), and R². Logistic regression models were evaluated using AIC, overall classification accuracy, sensitivity, specificity, and the area under the receiver operating characteristic curve (AUC).

2.9.2. Handling of Missing Data and Bias Control

No missing data were identified in the study dataset; therefore, no imputation procedures or complete-case exclusions were required. As all questionnaires were fully completed, weighting procedures and sensitivity analyses related to missing data were not necessary.

2.10. Data Cleaning and Variable Categorization

Prior to analysis, data were screened for inconsistencies and outliers. One record with an implausible age value (3 years) was excluded due to a data-entry error and a violation of the inclusion criterion (age ≥ 18 years). Age was categorized into groups to improve interpretability and statistical stability (18–44, 45–64, and ≥65 years).

Due to the small number of participants reporting COVID-19 hospitalization, self-reported COVID-19 infection history was recoded into two categories: never infected and previously infected (recovered), to ensure adequate cell counts for inferential analysis.

Depression, anxiety, and stress scores from the DASS-21 were analyzed both as continuous variables and as categorical severity levels according to established cutoff values. Categorical classification was used for descriptive and clinical interpretation, whereas continuous scores were used for inferential analyses.

3. Results

3.1. Sample Characteristics

The sociodemographic characteristics of the 260 participants are presented in Table 1. Overall, the sample was predominantly female and middle-aged, with most participants reporting secondary or higher education and being professionally active. Participants self-identified as either female or male; no participants selected the “other” gender category. Additional demographic and clinical characteristics are summarized in Table 1.

Table 1. Sociodemographic characteristics of the sample (n = 260).

	n	%
Gender
Male	104	40.0
Female	156	60.0
Age group (years)
18–44	117	45.0
45–64	89	34.2
≥65	54	20.8
Employment status
Professionally active	190	73.1
Not professionally active	70	26.9
Educational level
4th grade or less	22	8.5
6th grade	7	2.7
9th grade	26	10.0
High School	102	39.2
Higher Education	103	39.6
Marital status
Married/De facto union	157	60.4
Divorced	23	8.8
Single	69	26.5
Widowed	11	4.2

3.2. COVID-19 Status and Perceptions

The COVID-19-related and clinical characteristics of the sample are presented in Table 2. Most participants reported a previous SARS-CoV-2 infection, and almost half considered that the pandemic had negatively affected their mental health.

Table 2. COVID-19-related and clinical characteristics of the sample (n = 260).

Infection Status	n	%
Never infected	94	36.2
Recovered	166	63.8
Perceived mental health impact
No	139	53.5
Yes	121	46.5
Self-rated mental health after the pandemic
Poor	15	5.8
Fair	78	30.0
Good	115	44.2
Very good	52	20.0
Psychotropic medication initiated
No	232	89.2
Yes	28	10.8

3.3. Severity of Depression, Anxiety, and Stress

Table 3 presents the distribution of DASS-21 severity levels. The majority of participants fell within the normal range for depression (66.9%), anxiety (65.0%), and stress (74.6%). Nevertheless, a considerable proportion reported at least moderate symptom levels, particularly in the anxiety dimension.

Table 3. Distribution of DASS-21 severity levels (n = 260).

Variable	Normal, n (%)	Mild, n (%)	Moderate, n (%)	Severe, n (%)	Extremely Severe, n (%)
Depression	174 (66.9)	25 (9.6)	30 (11.5)	10 (3.8)	21 (8.1)
Anxiety	169 (65.0)	13 (5.0)	37 (14.2)	13 (5.0)	28 (10.8)
Stress	194 (74.6)	18 (6.9)	16 (6.2)	18 (6.9)	14 (5.4)

3.4. Inferential Analysis

As the continuous variables did not meet the assumption of normality, they are summarised as the mean and the median with the interquartile range (IQR), defined by the 25th and 75th percentiles. Group comparisons were performed using the Mann–Whitney U test or the Kruskal–Wallis test, as appropriate. p-values are reported in Table 4 and Table 5. For each comparison, the corresponding test statistic (U or H) and effect size (r for the Mann–Whitney test and ε² for the Kruskal–Wallis test) are presented in the text. For statistically significant comparisons, the estimated post hoc statistical power (1 − β_II) is also reported, using the most conservative parent distribution available in G*Power.

Table 4. Differences in levels of depression, anxiety, and stress according to self-reported COVID-19 infection history and sociodemographic characteristics.

Variable	Stress Mean/Median (IQR)	p (1 − βII)	Anxiety Mean (CI)	p (1 − βII)	Depression Mean (CI)	p (1 − βII)
Infection status
Never infected	8.00 (0–16)	0.713	2.00 (0–10)	0.181	7.55/2.00 (0–12)	0.388
Recovered	8.00 (2–14)		4.00 (0–10)		8.10/4.00 (0–10)
Age group (years)
18–44	6.00 (2–14)	0.012 (0.775)	4.00 (0–8)	0.003 (0.873)	6.44/2.00 (0–10)	0.003 (0.883)
44–64	6.00 (0–14)		4.00 (0.10)		6.72/4.00 (0–10)
≥65	12.00 (4–26)		8.00 (2–18)		13.00/11.00 (2–22)
Marital status
Married/De facto union	6.00 (2–14)	0.119	8.00 (0–10)	0.006 (0.870)	6.49/4.00 (0–12)	0.054
Divorced	4.00 (4–18)		8.00 (2–15)		7.31/8.00 (1–14)
Single	6.00 (0–14)		2.00 (0–8)		10.26/2.00 (0–8)
Widowed	26.00 (2–30)		18.00 (7–29)		20.18/28.00 (2–31)
Educational level
4th grade	12.00 (4–28)	0.126	8.00 (2–24)	0.008 (0.875)	16.45/11.00 (0–28)	0.047 (0.706)
6th grade	14.00 (9–21)		10.00 (3–19)		12.00/14.00 (3–19)
9th grade	8.00 (0–14)		6.00 (2–12)		8.34/6.00 (0–14)
High School	8.00 (2–14)		4.00 (0–10)		7.13/4.00 (0–11)
Higher Education	6.00 (0–14)		2.00 (0–9)		6.41/2.00 (0–10)
Employment status
Professionally active	6.00 (0–14)	<0.001 (0.988)	4.00 (0–8)	<0.001 (0.986)	6.32/2.00 (0–10)	<0.001 (0.969)
Not professionally active	12.00 (4–24)		7.00 (2–18)		12.20/8.00 (2–22)

Values are presented as mean/median (interquartile range). p = p-value of the Mann–Whitney U test (two groups) or of the Kruskal–Wallis test (three or more groups); the value in parentheses is the post hoc statistical power estimated for statistically significant comparisons.

Table 5. Differences in levels of depression, anxiety, and stress according to gender and COVID-19-related variables.

Variable	Stress Median (IQR)	p (1 − βII)	Anxiety Median (IQR)	p (1 − βII)	Depression Median (IQR)	p (1 − βII)
Gender
Male	4.00 (0–12)	<0.001 (0.999)	2.00 (0–6)	<0.001 (0.981)	2.00 (0–11)	0.080
Female	10.00 (3–16)		6.00 (2–12)		4.00 (0–13)
Perceived mental health impact
No	4.00 (0–8)	<0.001 (1.000)	2.00 (0–6)	<0.001 (1.000)	2.00 (0–5)	<0.001 (1.000)
Yes	14.00 (8–24)		8.00 (4–18)		10.00 (4–18)
Psychotropic medication initiated
No	6.00 (0–14)	<0.001 (1.000)	4.00 (0–9)	<0.001 (1.000)	3.00 (0–10)	<0.001 (1.000)
Yes	26.00 (14–34)		20.00 (8–27)		22.00 (8–34)
Self-rated mental health after the pandemic
Poor	32.00 (20–35)	<0.001 (1.000)	24.00 (19–34)	<0.001 (1.000)	32.00 (18–36)	<0.001 (1.000)
Fair	14.00 (8–22)		9.00 (4–16)		10.00 (2–18)
Good	0.00 (2–12)		4.00 (0–6)		2.00 (0–10)
Very good	0.00 (0–4)		0.00 (0–2)		0.00 (0–2)

Values are presented as mean/median (interquartile range). p = p-value of the Mann–Whitney U test (two groups) or of the Kruskal–Wallis test (three or more groups); the value in parentheses is the post hoc statistical power estimated for statistically significant comparisons.

3.5. Sociodemographic Differences

Women presented significantly higher anxiety scores (U(156,104) = 5945, r = 0.267, 1 − β_II = 0.981) and stress scores (U(156,104) = 5934.5, r = 0.268, 1 − β_II = 0.999) than men (both p < 0.001), whereas the difference in depression scores did not reach statistical significance (U(156,104) = 7092.5, r = 0.12, p = 0.080) (Table 5).

Age group differences were statistically significant for all three dimensions: stress (H(2) = 8.871, ε² = 0.034, p = 0.012; 1 − β_II = 0.775), anxiety (H(2) = 11.330, ε² = 0.044, p = 0.003; 1 − β_II = 0.876), and depression (H(2) = 11.503, ε² = 0.045, p = 0.003; 1 − β_II = 0.883). Participants aged ≥65 years presented the highest median scores for depression, anxiety, and stress.

Marital status was significantly associated with anxiety (H(3) = 12.558, ε² = 0.049, p = 0.006, 1 − β_II = 0.870), and educational level with anxiety (H(4) = 13.771, ε² = 0.053, p = 0.008, 1 − β_II = 0.875) and depression (H(4) = 9.627, ε² = 0.0372, p = 0.047, 1 − β_II = 0.706), whereas no significant differences were found for stress (H(3) = 5.860, ε² = 0.0226, p = 0.119) and (H(4) = 7.199, ε² = 0.0278, p = 0.126), respectively). Pairwise comparisons did not consistently identify significant differences between categories, likely owing to the small size of some subgroups.

Professionally inactive participants presented significantly higher depression (U(70,190) = 4804,5, r = 0.278, 1 − β_II = 0.969), anxiety (U(70,190) = 4662, r = 0.299, 1 − β_II = 0.986), and stress scores (U(70,190) = 4625, r = 0.305, 1 − β_II = 0.988) than professionally active participants (all p < 0.001). No significant differences were observed according to self-reported COVID-19 infection history (depression, p = 0.388, anxiety, p = 0.181 and stress, p = 0.713): depression (U(94,166) = 7309, r = 0.063), anxiety(U(94,166) = 7038, r = 0.098), and stress scores (U(94,166) = 7589.5, r = 0.027). For all statistically significant comparisons, the post hoc power exceeded 0.706 (Table 4).

3.6. COVID-19-Related Differences

The perceived psychological impact of the pandemic was strongly associated with all three emotional dimensions. Participants who reported a negative impact presented significantly higher depression, anxiety, and stress scores than those reporting no negative impact (all p < 0.001): depression (U(139,121) = 4207, r = 0.500, 1 − β_II = 1.000), anxiety (U(139,121) = 4305, r = 0.488, 1 − β_II = 1.000), and stress scores (U(139,121) = 3770.5, r = 0.552, 1 − β_II = 1.000).

Similarly, participants who initiated psychotropic medication after the pandemic presented markedly higher symptom scores across all dimensions (all p < 0.001)): depression (U(232,28) = 1255.5, r = 0.614, 1 − β_II = 1.000), anxiety (U(232,28) = 1082.5, r = 0.667, 1 − β_II = 1.000), and stress scores (U(232,28) = 852.5, r = 0.867, 1 − β_II = 1.000).

Self-rated mental health after the pandemic was significantly associated with all DASS-21 dimensions (all p < 0.001): stress (H(2) = 47.721, ε² = 0.184, 1 − β_II = 1.000) anxiety (H(2) = 47.49, ε² = 0.183, 1 − β_II = 1.000), depression, (H(2) = 41.585, ε² = 0.161, 1 − β_II = 1.000). A clear gradient was observed: poorer self-rated mental health corresponded to higher levels of depression, anxiety, and stress (Table 5).

3.7. Factors Independently Associated with Mental Health Outcomes

Multivariable generalized linear models were used to assess the independent contribution of each variable to the mental health outcomes. Table 6 presents the models including sociodemographic variables only.

Table 6. Multivariable generalized linear models of the association between sociodemographic characteristics and depression, anxiety, stress, and perceived impact on mental health.

Variables	Depression Score 1 β/exp(β) [95% CI]	Anxiety Score 1 β/exp(β) [95% CI]	Stress Score 1 β/exp(β) [95% CI]	Perceived Impact on Mental Health 2 β/aOR [95% CI]
Age group (45–64 and ≥65 vs. <45 years)	—	—	—	—
Gender (male vs. female)	−0.48/0.62 * [0.42–0.92]	−0.72/0.49 *** [0.34–0.71]	−0.48/0.61 ** [0.45–0.84]	−0.75/0.47 ** [0.28–0.80]
Employment status (active vs. inactive)	−0.55/0.58 * [0.37–0.89]	−0.50/0.61 * [0.40–0.90]	−0.57/0.56 ** [0.40–0.79]	−1.22/0.30 *** [0.16–0.53]
Education: 12th grade (vs. ≤9th grade)	−0.40/0.67 [0.40–1.12]	−0.49/0.61 * [0.37–0.99]	—	—
Education: university degree (vs. ≤9th grade)	−0.53/0.59 * [0.34–0.99]	−0.50/0.57 * [0.35–0.92]	—	—
Model fit
AIC	1547.03	1511.70	1722.02	340.73
Residual deviance	288.78	291.41	306.56	—
RMSE	9.52	8.44	9.97	—
R2	0.10	0.12	0.09	—
Overall classification accuracy	—	—	—	62.69%
Sensitivity/specificity	—	—	—	38.84%/83.45%
AUC	—	—	—	0.667

¹ Negative binomial regression with a log link; β and exp(β) with 95% confidence interval. ² Logistic regression with a logit link; β and adjusted odds ratio (aOR) with 95% confidence interval. Reference categories: age < 45 years, female gender, professionally inactive, education up to the 9th grade. Predictors not retained by the stepwise procedure are indicated by an em dash. AIC = Akaike Information Criterion; AUC = area under the receiver operating characteristic curve; CI = confidence interval; RMSE = root mean squared error. * p ≤ 0.05; ** p ≤ 0.01; *** p ≤ 0.001.

Gender was consistently associated with all mental health outcomes. Compared with female participants, male participants showed significantly lower expected depression scores [exp(β) = 0.62, 95% CI: 0.42–0.92, p < 0.05], anxiety scores [exp(β) = 0.49, 95% CI: 0.34–0.71, p < 0.001], and stress scores [exp(β) = 0.61, 95% CI: 0.45–0.84, p < 0.01]. Male participants also had lower odds of reporting a perceived negative impact on mental health [aOR = 0.47, 95% CI: 0.28–0.80, p < 0.01].

Employment status was likewise associated with all outcomes. Being professionally active was associated with lower expected depression [exp(β) = 0.58, 95% CI: 0.37–0.89, p < 0.05], anxiety [exp(β) = 0.61, 95% CI: 0.40–0.90, p < 0.05], and stress scores [exp(β) = 0.56, 95% CI: 0.40–0.79, p < 0.01], as well as with substantially lower odds of reporting a negative perceived impact on mental health [aOR = 0.30, 95% CI: 0.16–0.53, p < 0.001].

Educational level was associated primarily with depression and anxiety. Compared with participants with education up to the 9th grade, those with a university degree or equivalent showed lower expected depression scores [exp(β) = 0.59, 95% CI: 0.34–0.99, p < 0.05]. Both completion of the 12th grade [exp(β) = 0.61, 95% CI: 0.37–0.99, p < 0.05] and university education [exp(β) = 0.57, 95% CI: 0.35–0.92, p < 0.05] were associated with lower expected anxiety scores.

The sociodemographic models showed a modest explanatory capacity. The depression, anxiety, and stress models presented R² values of 0.10, 0.12, and 0.09, respectively, with corresponding AIC values of 1547.03, 1511.70, and 1722.02. The logistic regression model for perceived impact on mental health achieved an overall classification accuracy of 62.69%, with a sensitivity of 38.84%, a specificity of 83.45%, and an AUC of 0.667.

It should be noted that some associations that were statistically significant in the sociodemographic models were no longer retained, or no longer reached statistical significance, after the inclusion of COVID-19-related and health-related variables. This does not necessarily indicate that the original associations were absent; rather, the inclusion of additional correlated explanatory variables allowed the models to account for shared variation among predictors. Consequently, the independent contribution of individual variables changed after mutual adjustment, and the stepwise selection procedure retained the predictors providing the best overall model fit.

After the inclusion of self-reported COVID-19 infection history, psychotropic medication use, and self-rated health after COVID-19 in the multivariable models, several additional associations with mental health outcomes were observed (Table 7).

Table 7. Multivariable generalized linear models of the association between sociodemographic, COVID-19-related, and health-related characteristics and depression, anxiety, stress, and perceived impact on mental health.

Variables	Depression Score 1 β/exp(β) [95% CI]	Anxiety Score 1 β/exp(β) [95% CI]	Stress Score 1 β/exp(β) [95% CI]	Perceived Impact on Mental Health 2 β/aOR [95% CI]
Age group: 45–64 (vs. <45 years)	—	—	−0.05/0.95	—
Age group: ≥65 (vs. <45 years)	—	—	+0.42/1.52 *	—
Gender (male vs. female)	—	−0.45/0.64 ** [0.46–0.89]	−0.46/0.63 **	−0.72/0.49 * [0.28–0.84]
Employment status (active vs. inactive)	−0.36/0.70 ● [0.47–1.02]	−0.30/0.74 [0.51–1.05]	—	−1.23/0.29 *** [0.16–0.53]
Education: 12th grade (vs. ≤9th grade)	—	−0.46/0.63 * [0.42–0.96]	—	—
Education: university degree (vs. ≤9th grade)	—	−0.60/0.55 ** [0.36–0.83]	—	—
Self-reported COVID-19 infection history (yes vs. no)	—	+0.32/1.37 ● [0.99–1.90]	—	—
Psychotropic medication after COVID-19 (yes vs. no)	+0.55/1.74 ● [0.97–3.31]	+0.60/1.83 * [1.10–3.15]	+0.65/1.92 **	+2.10/8.16 *** [2.93–29.14]
Self-rated health: fair (vs. poor)	−0.49/0.61 [0.26–1.30]	−0.41/0.66 [0.32–1.28]	−0.11/0.89 **	—
Self-rated health: good (vs. poor)	−1.08/0.34 ** [0.14–0.74]	−0.93/0.40 * [0.19–0.79]	−0.63/0.53	—
Self-rated health: very good (vs. poor)	−1.98/0.14 *** [0.05–0.32]	−2.05/0.13 *** [0.06–0.27]	−1.55/0.53	—
Model fit
AIC	1504.01	1455.56	1665.72	324.23
Residual deviance	294.54	298.91	315.10	—
RMSE	7.63	7.26	7.99	—
R2	0.42	0.40	0.43	—
Overall classification accuracy	—	—	—	67.31%
Sensitivity/specificity	—	—	—	51.23%/81.29%
AUC	—	—	—	0.720

¹ Negative binomial regression with a log link; β and exp(β) with 95% confidence interval. ² Logistic regression with a logit link; β and adjusted odds ratio (aOR) with 95% confidence interval. Reference categories: age < 45 years, female gender, professionally inactive, education up to the 9th grade, no previous COVID-19 infection, no psychotropic medication, poor self-rated health. Predictors not retained by the stepwise procedure are indicated by an em dash. ● p ≤ 0.10; * p ≤ 0.05; ** p ≤ 0.01; *** p ≤ 0.001.

Gender remained significantly associated with anxiety and stress. Compared with female participants, male participants showed lower expected anxiety scores [exp(β) = 0.64, 95% CI: 0.46–0.89, p < 0.01] and lower expected stress scores [exp(β) = 0.63, p < 0.01], as well as lower odds of reporting a negative perceived impact on mental health [aOR = 0.49, 95% CI: 0.28–0.84, p < 0.05].

Employment remained associated with depression, although the association was only marginally significant: professionally active participants showed lower expected depression scores [exp(β) = 0.70, 95% CI: 0.47–1.02, p < 0.10]. Active employment remained strongly associated with lower odds of a negative perceived impact on mental health [aOR = 0.29, 95% CI: 0.16–0.53, p < 0.001].

Educational level remained associated with anxiety. Compared with participants with education up to the 9th grade, those who completed the 12th grade [exp(β) = 0.63, 95% CI: 0.42–0.96, p < 0.05] and those with a university degree or equivalent [exp(β) = 0.55, 95% CI: 0.36–0.83, p < 0.01] showed lower expected anxiety scores.

Self-reported COVID-19 infection history showed only a marginal association with anxiety, with participants reporting a previous infection presenting higher expected anxiety scores [exp(β) = 1.37, 95% CI: 0.99–1.90, p < 0.10].

Psychotropic medication use after COVID-19 was consistently associated with poorer mental health outcomes. Participants reporting psychotropic medication use showed marginally higher expected depression scores [exp(β) = 1.74, 95% CI: 0.97–3.31, p < 0.10], significantly higher anxiety scores [exp(β) = 1.83, 95% CI: 1.10–3.15, p < 0.05], and higher stress scores [exp(β) = 1.92, p < 0.01]. Psychotropic medication use was also strongly associated with higher odds of reporting a negative perceived impact on mental health [aOR = 8.16, 95% CI: 2.93–29.14, p < 0.001].

Self-rated health after COVID-19 was strongly associated with depression and anxiety. Compared with participants reporting poor health, those reporting good health showed substantially lower expected depression [exp(β) = 0.34, 95% CI: 0.14–0.74, p < 0.01] and anxiety scores [exp(β) = 0.40, 95% CI: 0.19–0.79, p < 0.05], and those reporting very good health showed even lower expected depression [exp(β) = 0.14, 95% CI: 0.05–0.32, p < 0.001] and anxiety scores [exp(β) = 0.13, 95% CI: 0.06–0.27, p < 0.001].

The inclusion of COVID-19-related and health-related variables substantially improved the explanatory capacity of the models. The R² values increased to 0.42, 0.40, and 0.43 for depression, anxiety, and stress, respectively, and the corresponding AIC values decreased to 1504.01, 1455.56, and 1665.72. The logistic regression model for perceived impact on mental health achieved an overall classification accuracy of 67.31%, with a sensitivity of 51.23%, a specificity of 81.29%, and an AUC of 0.720.

4. Discussion

The present study aimed to characterize levels of depression, anxiety, and stress in a primary care population in the post-COVID-19 pandemic context and to examine their associations with sociodemographic and pandemic-related variables. Overall, the findings indicate that although the majority of participants fell within the normal range of symptom severity, a substantial proportion reported clinically relevant levels of psychological distress, particularly in the domains of anxiety and depression.

4.1. Prevalence and Clinical Relevance of Emotional Distress

Approximately one-third of the sample presented at least moderate levels of depressive or anxiety symptoms, and roughly one-quarter reported moderate or higher stress levels. These findings are consistent with international literature documenting sustained psychological burden beyond the acute phase of the pandemic [2,7]. While symptom severity in this study was not universally high, the proportion of participants with moderate-to-extremely severe symptoms remains clinically significant in a primary care setting, where early detection and intervention are crucial.

The presence of emotional symptoms in the post-pandemic period indicates that a substantial proportion of primary care users continue to experience psychological distress. However, because of the cross-sectional design, it cannot be determined whether these symptoms are direct consequences of the COVID-19 pandemic or reflect broader trends in population mental health. The observed associations may also be influenced by other social, economic, or individual factors not assessed in the present study.

Moreover, psychological symptoms are multifactorial and are influenced by a complex interplay of biological, psychological, social, and economic determinants. Although this study examined several sociodemographic and COVID-19-related variables, other important contextual factors, including financial strain, chronic physical illness, bereavement, and social support, were not measured. These unmeasured factors may have influenced the observed associations and should be incorporated into future research.

Given that data collection occurred several years after the acute phase of the COVID-19 pandemic, the observed symptoms should not be interpreted as evidence of persistent psychological consequences of SARS-CoV-2 infection or the pandemic itself. Rather, the findings describe the mental health status of a primary care population in the post-pandemic period and identify factors associated with psychological distress at the time of assessment. Other social, economic, health-related, and individual factors occurring after the pandemic may also have contributed to the observed symptom levels.

Although previous psychiatric disorders were recorded as a participant characteristic, the present study was not designed to evaluate their independent contribution to psychological symptom severity. Future studies using multivariable analyses should investigate the extent to which pre-existing psychiatric conditions influence post-pandemic psychological distress.

These findings are consistent with the growing body of literature describing post-acute and long-term psychosocial sequelae reported following the COVID-19 pandemic. However, because of the cross-sectional design of the present study, our results should be interpreted as associations observed in the post-pandemic period rather than evidence that the pandemic caused or sustained the reported psychological symptoms.

4.2. Sociodemographic Vulnerability Factors

Significant differences were observed across several sociodemographic variables. Women reported higher anxiety and stress levels than men, a finding widely documented in mental health research. Biological vulnerability, differential stress exposure, and gender-related social roles may partly explain this pattern [8,9].

An unexpected finding was that participants aged ≥65 years presented higher depression, anxiety, and stress symptom scores than younger participants. This contrasts with many studies conducted during the acute phase of the COVID-19 pandemic, which frequently reported greater psychological distress among younger adults, potentially related to educational disruption, employment insecurity, and social restrictions. Several factors may explain the discrepancy. First, the present study was conducted in the post-pandemic period, and psychological determinants may differ from those observed during the early stages of the pandemic. Second, our sample consisted of primary care users rather than a community-based population; older adults attending a Family Health Unit may have a greater burden of chronic medical conditions, functional limitations, healthcare dependence, social isolation, or bereavement, all of which are associated with psychological distress. Third, because these contextual factors were not directly measured, residual confounding cannot be excluded. Therefore, this finding should be interpreted cautiously and viewed as hypothesis-generating rather than evidence that older age is associated with higher psychological symptoms in the general population [10,11].

Professionally inactive individuals demonstrated significantly higher levels of emotional distress across all dimensions. Employment status is closely linked to financial stability, social engagement, and a sense of purpose, and its absence may exacerbate psychological vulnerability. The moderate effect sizes observed highlight the practical relevance of this association [12,13].

Although marital status and educational level showed overall statistical differences, post hoc comparisons were less consistent, likely due to smaller subgroup sizes. Nonetheless, lower educational attainment was associated with higher depressive symptomatology, aligning with established evidence linking socioeconomic disadvantage to mental health risk [14,15].

These unadjusted associations were largely confirmed in the multivariable models. Male gender and professional activity remained independently associated with lower depression, anxiety, and stress scores and with lower odds of reporting a negative perceived impact on mental health, whereas higher educational attainment remained independently associated with lower anxiety scores. After the inclusion of COVID-19-related and health-related variables, however, the explanatory capacity of the models increased markedly (R² from approximately 0.10 to above 0.40), and psychotropic medication use and self-rated health after COVID-19 emerged as the strongest correlates of psychological distress. This pattern suggests that current perceived health status and ongoing pharmacological treatment carry more information about symptom burden than sociodemographic characteristics alone, and that part of the sociodemographic effect is shared with these health-related indicators.

4.3. COVID-19 Infection Versus Psychological Impact

One of the most noteworthy findings of this study is the absence of significant differences in emotional distress between participants who had been infected with COVID-19 and those who had not. Differences between the two groups were not statistically significant, and self-reported infection history was not retained as a predictor of depression or stress in the multivariable models, suggesting that the clinical experience of infection per se was not a primary associated factor of psychological symptom levels in this sample [16].

In contrast, the perceived psychological impact of the pandemic demonstrated strong and clinically meaningful associations with depressive, anxiety, and stress symptoms. These associations were of large magnitude and remained evident after adjustment for sociodemographic, COVID-19-related, and health-related variables. Similarly, participants who reported initiating psychotropic medication during or after the pandemic presented significantly higher depression, anxiety, and stress symptom scores. This finding should not be interpreted as indicating that psychotropic medication increases psychological symptoms. Rather, it most likely reflects confounding by indication, whereby individuals with more severe or persistent psychological symptoms are more likely to receive pharmacological treatment. Consequently, psychotropic medication initiation should be viewed as a marker of greater underlying psychological distress rather than an independent risk factor [17,18]. Because of the cross-sectional design, the temporal relationship between symptom severity and medication initiation cannot be established. Longitudinal studies are needed to determine whether symptom trajectories change following treatment initiation.

It should also be considered that COVID-19 infection history and the perceived psychological impact of the pandemic were self-reported and assessed several years after the acute phase of the pandemic. These measures may therefore be subject to recall and reporting bias, which should be considered when interpreting the observed associations.

These results suggest that participants’ subjective appraisal of the pandemic was more strongly associated with emotional well-being than their self-reported infection status. This distinction reinforces the importance of considering psychological and social dimensions, rather than focusing solely on biomedical indicators.

4.4. Self-Perceived Mental Health as a Key Indicator

Self-rated mental health after the pandemic emerged as one of the strongest correlates of DASS-21 scores, with strong associations across all dimensions that persisted in the multivariable models. A clear gradient was observed: poorer self-perceived mental health corresponded to progressively higher levels of depressive, anxiety, and stress symptoms.

Although 46.5% of participants reported that the COVID-19 pandemic had negatively affected their mental health, a considerably smaller proportion rated their current mental health as poor or fair. This apparent discrepancy likely reflects the different constructs assessed. The perceived psychological impact variable captured participants’ retrospective appraisal of whether the pandemic had adversely affected their mental health at any point during the pandemic, whereas the self-rated mental health item assessed their current mental health status in the post-pandemic period. Thus, participants may acknowledge having experienced psychological distress during the pandemic while perceiving that their mental health had subsequently improved.

This finding supports the value of simple self-assessment questions in primary care practice. Self-rated mental health may serve as a rapid screening indicator that identifies individuals at higher risk of clinically relevant symptomatology [19].

4.5. Clinical and Public Health Implications

From a primary care perspective, the findings underscore the importance of routine mental health screening in the post-pandemic context. The substantial proportion of patients presenting moderate or higher symptom levels suggests that psychological distress remains an ongoing concern [5].

Importantly, the data indicate that perceived psychological burden and functional impact may be more critical targets for intervention than infection history alone. Primary care teams should therefore consider incorporating structured mental health assessment tools, particularly for women, older adults, professionally inactive individuals, and those reporting negative psychological impact of the pandemic.

Although the study examined psychological well-being in the context of the COVID-19 pandemic, its cross-sectional design and the timing of data collection several years after the acute phase of the pandemic preclude any causal inference regarding the origin or persistence of depression, anxiety, and stress symptoms. The observed associations should therefore be interpreted as reflecting participants’ current mental health status and self-reported experiences within the post-pandemic context rather than direct effects of the pandemic itself.

Although several associations reached statistical significance, their clinical relevance should also be considered. The findings suggest that certain patient characteristics—particularly poorer self-rated mental health, a history of psychiatric disorders, psychotropic medication use, and a negative perceived psychological experience related to the pandemic—may help primary care professionals identify individuals with an increased burden of depressive, anxiety, and stress symptoms. While these variables should not be interpreted as causal or independent risk factors, they may serve as practical indicators for targeted mental health screening and timely referral or intervention within primary care. Given the accessibility of primary care services, incorporating routine assessment of psychological symptoms in patients presenting these characteristics may facilitate earlier recognition and management of emotional distress.

4.6. Strengths and Limitations

This study has several strengths, including a relatively large sample size, the use of validated psychometric instruments, the use of non-parametric methods appropriate to the observed data distributions, the reporting of effect sizes and post hoc statistical power alongside significance testing, and the use of multivariable generalized linear models, which allow the independent contribution of each variable to be evaluated.

However, certain limitations must be acknowledged. First, its cross-sectional design precludes causal inference and does not permit determination of the temporal direction of the observed associations. Consequently, reverse causality cannot be excluded. For example, participants with greater psychological distress may have been more likely to initiate psychotropic medication or to perceive the pandemic as having negatively affected their mental health, rather than these factors leading to higher symptom levels.

The study used a convenience sample of adults attending a single Family Health Unit. Consequently, participants were not randomly selected, and the findings should not be considered representative of the general Portuguese population or all primary care users. The results should therefore be interpreted as applying to this study population and confirmed in larger, multicentre studies employing probability-based sampling.

Participants were recruited from adults attending a Family Health Unit, which may have introduced selection bias. Individuals seeking primary healthcare are more likely to have underlying medical conditions, greater healthcare utilization, or health-related concerns than the general population, factors that may also be associated with psychological symptoms. Consequently, the findings may not be representative of community-dwelling adults who do not regularly access primary care services. The results should therefore be interpreted within the context of a primary care population.

An additional limitation is that unemployed individuals, retirees, and students were combined into a single “professionally inactive” category because of small subgroup sizes. Although this approach improved statistical stability, these groups may differ in important demographic and socioeconomic characteristics, particularly age, which could have influenced the observed associations. Future studies with larger samples should examine these groups separately.

An additional limitation relates to the timing and measurement of the COVID-19-related variables. Data were collected during the post-pandemic period, several years after the acute phase of the COVID-19 pandemic. Consequently, self-reported COVID-19 infection history and the perceived psychological impact of the pandemic may have been influenced by recall bias and subjective interpretation. Furthermore, no objective confirmation of infection status or longitudinal assessment of mental health was available. Therefore, these findings indicate that symptoms of depression, anxiety, and stress remain common among primary care users in the post-pandemic period. However, because no pre-pandemic baseline data were available, it cannot be determined whether these symptom levels represent persistent effects of the pandemic or reflect broader patterns of mental health in this population.

Furthermore, because no pre-pandemic data, comparison group, or longitudinal follow-up were available, it was not possible to determine whether the observed psychological symptoms resulted from the COVID-19 pandemic or reflected broader secular trends in population mental health.

Another important limitation is that COVID-19 infection history was based exclusively on self-report and was not confirmed by laboratory testing or medical records. Consequently, participants classified as reporting no previous COVID-19 infection may have experienced asymptomatic, undiagnosed, or unrecognized infections. This potential misclassification may have attenuated differences between groups and should be considered when interpreting the absence of associations between self-reported infection history and psychological outcomes.

Because participants were recruited while attending a primary care consultation, it is possible that individuals seeking healthcare differed from the general population in terms of health status or psychological well-being. Although participation was voluntary and independent of clinical care, this recruitment strategy may have introduced selection bias and limits the generalizability of the findings.

An additional limitation is the absence of pre-pandemic baseline data or longitudinal follow-up. Consequently, it is not possible to determine whether the observed levels of depression, anxiety, and stress symptoms represent changes attributable to the COVID-19 pandemic or reflect pre-existing patterns of psychological distress in this population. The findings should therefore be interpreted as a cross-sectional description of mental health in the post-pandemic period rather than evidence of persistent pandemic-related effects.

The study was conducted at a single-Family Health Unit in Northern Portugal, which may limit the generalizability of the findings to other primary care settings or to the wider Portuguese population. Differences in demographic characteristics, socioeconomic conditions, healthcare organization, and access to mental health services across regions may influence psychological symptom profiles. Therefore, the findings should be interpreted within the context of the study population. Future multicenter studies including geographically and socioeconomically diverse primary care populations are warranted to enhance external validity.

Furthermore, the observed association between psychotropic medication initiation and higher DASS-21 scores is susceptible to confounding by indication and should not be interpreted as evidence of a causal relationship.

Another limitation is that several contextual factors known to influence psychological well-being were not assessed. Variables such as socioeconomic hardship, chronic medical comorbidities, bereavement, social support, and other major life stressors may have contributed to symptoms of depression, anxiety, and stress independently of the COVID-19 pandemic. The absence of these variables limits the ability to account for residual confounding and should be considered when interpreting the observed associations. Future studies should include a broader range of psychosocial and clinical determinants to better characterize factors associated with psychological distress in primary care populations.

Future research employing longitudinal designs would be valuable for determining whether observed symptom levels reflect psychological symptoms observed in the post-pandemic period or broader trends in population mental health.

5. Conclusions

In this single-center sample of primary care users in Northern Portugal, a substantial proportion of participants presented clinically relevant levels of depression, anxiety, and stress as measured by the DASS-21. No significant associations were observed between self-reported history of COVID-19 infection and DASS-21 scores. Because infection history relied on participant self-report and may have been subject to misclassification, these findings should be interpreted with caution. In contrast, participants who reported a negative psychological impact of the COVID-19 pandemic had higher depression, anxiety, and stress scores than those who did not report such an impact. Similarly, poorer self-rated mental health, initiation of psychotropic medication after the pandemic, female sex, older age, and professional inactivity were associated with higher symptom levels. In multivariable generalized linear models, female gender, professional inactivity, lower educational attainment, psychotropic medication use, and poorer self-rated health after COVID-19 remained independently associated with higher symptom scores.

These findings should be interpreted in light of the cross-sectional design, which does not permit causal inference or conclusions regarding the psychological consequences of the COVID-19 pandemic. Nevertheless, the results describe the mental health status of this primary care population during the post-pandemic period and should not be interpreted as evidence that the observed symptoms were caused by the COVID-19 pandemic. They nonetheless highlight the value of routine mental health assessment in primary care and support further longitudinal, multicentre studies to better understand the determinants of psychological distress.