The Medium-Term Psychosocial Impact of the 2021 Floods in Belgium: A Survey-Based Study

Abstract

Background: This study investigates the medium-term psychosocial impacts of the 2021 floods in Belgium, which caused fatalities and considerable infrastructural damage. Given similar events’ significant impacts on psychosocial well-being, this study seeks to answer three questions: whether there are medium-term (two years and further) effects on residents’ psychosocial well-being, whether demographic variables influence these effects, and how flood exposure impacts psychosocial well-being. Methods: We collected data in affected municipalities through an online survey, assessing demographic variables (e.g., age, gender, education, SES), flood exposure (e.g., being physically hurt, being faced with financial difficulties), and psychosocial well-being, employing two validated instruments for quantitative evaluation: the RAND-36 and the Traumatic Exposure Severity Scale (TESS). Results: The sample included 114 participants, with 54% reporting a deterioration in their psychosocial well-being after the floods. Additionally, over 50% mentioned the psychosocial impact of the floods. SES was the only significant demographic variable impacting psychosocial well-being, with lower SES linked to higher deterioration. Financial difficulties generated by the floods were the only considerable exposure factor. Furthermore, 22% discussed being unhappy with the organized response measures. Due to the sample size, confounding effects could not be checked. Conclusions: This study found a medium-term effect of the 2021 floods on psychosocial well-being, highlighting the need for policy adaptations focused on post-disaster psychosocial support. With lower SES and financial difficulties as risk factors, one needs to design policies tailored to these vulnerable groups. With climate change expected to increase flood events, context-specific policies are essential to boost resilience.

1. Introduction

In July 2021, parts of Belgium, Germany, and the Netherlands were hit by extreme rainfall, causing flash floods. In Belgium, the province of Liège suffered the most severe impact, counting 41 human losses and more than 100,000 affected inhabitants [1,2], as shown in Figure 1. Authorities in Belgium were not fully prepared for a flood of this magnitude, causing the emergency response to be difficult and slow, potentially intensifying the impact of the events on the population [3].

In mainland Northern Europe, floods are the most common natural disaster, and scholars expect their occurrence to rise in severity and frequency as climate change becomes more tangible [8,9]. The Joint Research Center of the European Environment Agency has developed an illustration (see Appendix A) that shows the relative change in expected damage for Western Europe between the current period and future scenarios ranging from 2071 to 2100 [10].

We may expect that this will also impact the human health of affected populations in Europe. Between 2000 and 2017, floods took the lives of 2000 people and affected more than 8 million in the Europe World Health Organization (WHO) region; the damages and losses cost over EUR 70 billion [9]. Moreover, the highest health burden generated by floods in high-income settings is related to mental health and psychosocial well-being, pointing to the relevance of this issue in the region [11]. What is more, the WHO’s broad range of climate change-related events are associated with the exacerbation or generation of mental health conditions [12].

1.1. Context and Concepts

“Mental health” and “psychosocial well-being” are often used interchangeably; however, some researchers prefer the term “psychosocial well-being” due to the strong association of “mental health” with psychopathologies. “Psychosocial well-being” includes “feeling positive emotions, feeling accomplished, being optimistic about the future, etc.” [13,14], which aligns with the WHO’s definition of health [15]. This study focuses on psychosocial well-being, aiming to assess the broader dimensions of psychosocial quality of life. Within the existing literature, the related concept of psychological well-being is understood as a distinct yet interconnected construct. Psychological well-being encompasses broader dimensions such as life satisfaction, resilience, and emotional regulation, while mental health is more often associated with the absence of mental illness or distress. Drawing from this literature, we reflect on how psychological well-being represents a proactive and positive framing of mental health, emphasizing resilience-building while also acknowledging how psychological well-being is shaped by environmental stressors such as natural disasters. Importantly, psychological well-being can diminish subsequent disasters due to disruptions in life satisfaction or resilience, even when individuals do not meet the criteria for clinical mental health conditions. This distinction is critical for interpreting and situating our findings within broader frameworks that distinguish well-being from mental health, while still recognizing overlapping elements. Finally, this study draws on the alignment between psychological well-being and holistic approaches to health and resilience-building, particularly in the context of disaster events such as the flooding of 2021. These conceptual underpinnings provide a theoretical basis for our study.

Despite acknowledging the mental health risks of climate change, the Joint Programming Initiative “Connecting Climate Knowledge for Europe” (JPI-Climate), an initiative of European member states and associated members to align national programs by jointly coordinating their climate research and funding new transnational research activities, reflected how people and communities in the region recognize the catastrophic effects of climate change but believe that these effects will be more severe elsewhere [16].

Most studies evaluate the impact on psychosocial well-being shortly after exposure (up to two years post-event), but knowledge on how this impact develops in the medium term (two years and further) is limited [17,18,19]. Factors such as mourning or financial worries can have medium- and long-term impacts on one’s psychosocial well-being [17]. Hwong and colleagues (2022) found that quantitative assessments of mental health impacts of climate-exacerbated disasters are usually conducted shortly after exposure, leaving long-term effects understudied [19]. Stanke and colleagues (2012) report that most studies conduct mental health impacts of floods 6–24 months post-event [18]. This aligns with Fernandez et al. (2015), who found few studies investigating medium- to long-term psychosocial impacts, but with those few finding that adverse effects persist 3–5 years after exposure [11,20], illustrating the need for additional research on the impact of extreme weather events on psychosocial well-being [21]. This overall lack of evidence is problematic, as we need to thoroughly understand the pathways that influence climate-related events on mental health in the short to long term, build more resilient societies, and appropriately adapt health interventions and psychosocial prevention strategies [22].

1.2. Objective and Research Questions

Studying the 2021 floods in the Vesdre Valley in Belgium within the context of summer floods in Europe is particularly relevant in the above-stated context. To the authors’ knowledge, no studies or official government reports have been published that specifically examine the psychosocial impact of these floods in the Belgian affected areas. This study, therefore, represents one of the first academic efforts to investigate the psychosocial consequences of the 2021 floods in this region, contributing to a better understanding of the medium-term effects on affected populations. Concretely, this study aims to evaluate whether there is a medium-term (24–36 months) impact of the 2021 floods on the psychosocial well-being of inhabitants of affected municipalities. This research strives to answer the following questions:

• Do affected persons experience an impact on their psychosocial well-being more than two years after exposure?
• Do certain demographic variables (gender, SES, etc.) influence the psychosocial well-being impacts experienced?
• Does the level of exposure to the floods (e.g., being physically hurt during the disaster, having lost property, being displaced) influence the impact on psychosocial well-being?

2. Methodology

2.1. Study Design

Data were collected through a quantitative research design using an online survey in French. The study received ethics approval by the commission for medical ethics of Ghent University Hospital on 22 October 2022 (with approval number ONZ-2022-0107). The survey consisted of three sections: (1) demographic data collection, (2) the level of exposure to floods, and (3) psychosocial well-being. In terms of study population and recruitment, any individual who was a resident of an impacted municipality during the floods was considered an eligible participant. Four methods of recruitment were used:

• Contacting local networks, e.g., community organizations and Facebook groups that were willing to share the link to the survey;
• Reaching out to municipalities that could spread the survey link through their communication channels (social media pages, municipal magazine);
• Requesting that healthcare providers in the area who are UNU Climate Resilience Initiative contacts from previous research share the survey in their patient network;
• The snowball effect, where recruited people contacted others.

Respondents were not offered compensation for their participation. Responses were collected in a period between 24 and 36 months after the July 2021 floods, with most responses being collected 30 months after exposure. The collected data were anonymized, with only researcher NDM having access to identifiable data, to protect the privacy of the respondents. We collected 205 initial responses, of which 114 were used in the analysis. Other responses were incomplete.

2.2. Variables and Measures

Demographic variables and exposure severity are this study’s independent variables; psychosocial well-being is the dependent variable. We added the questionnaire in Appendix B. We selected the variables based on their relevance, as determined by the literature. The survey component that assesses severity of exposure was based on the Traumatic Exposure Severity Scale (TESS), developed to evaluate someone’s disaster exposure. The TESS initially tested earthquake exposure, but the authors indicate that the questionnaire can be adapted to fit other disaster types [23]. The TESS scale used in this study consists of the following items: Was your home damaged in the floods? Did you lose movable goods? Did you need shelter, food, or water aid? Did you suffer financial difficulties? Did you have to relocate due to structural damage? Were you or your loved ones physically injured? Did you lose a loved one? Each item has the same response options: Yes or No. Items are analyzed separately, not combined into a composite score. Appendix B shows a full breakdown of the scale items,

To collect data on the respondents’ psychosocial well-being, we used the validated Mental Health Component (MHC) of the RAND-36 questionnaire. The MHC consists of the following blocks: mental health, vitality, social functioning, and limited role functioning due to emotional concerns [24,25]. The Cronbach’s alpha is 0.53. With these blocks, this questionnaire allows a thorough scope of the current psychosocial well-being of the respondents. For this research, we used a validated French translation [25,26]. We scored the responses to the RAND-36 following the steps defined in the literature, resulting in an MHC score out of 100 for each participant, with a higher score corresponding to higher well-being [24,27]. In addition to the MHC scale, the survey included a 5-point scale to assess whether the respondents’ psychosocial well-being had improved, deteriorated, or remained the same as before the floods. We used the following scoring: 1—strong deterioration, 2—slight deterioration, 3—no change, 4—slight improvement, and 5—strong improvement, meaning that a higher score corresponds with higher well-being. We refer to this variable as the psychosocial change (PSC) score to indicate the floods’ direct psychosocial impact. Finally, respondents could leave an open comment, resulting in qualitative data. They were prompted to do so with the question “Is there anything you would like to add?”

2.3. Data Analysis

For the quantitative analysis part, we conducted bivariate analyses to evaluate the impact of specific demographics on the PSC score and the MHC score. Using Levene’s test, the assumption of homogeneity of variances for a t-test was not met. Instead, a Mann–Whitney U-test was used to assess the difference between PSC scores for men and women. (A third gender category, “other”, was included in the questionnaire. However, none of the participants selected this option. Respondents also had the option not to share their gender, which only one person did. Therefore, the researchers performed a t-test with two independent samples, including only men and women.) In addition, ANOVA analyses were conducted to compare the outcomes of psychosocial impact between different demographic groups. Visual distribution inspection was used to check for normality of the distributions. A visual box plot analysis was used to check for outliers, and Levene’s test was used to check for homogeneity of variance. Since the variable education did not meet the assumption of homogeneity of variance, we used a Kruskal–Wallis test. We performed multiple linear regression to evaluate differences in the PSC scores between respondents having experienced different exposure levels. The assumptions of linearity, homoscedasticity, and normality of residuals in this test were checked by visual inspection using scatter plots. In addition, multicollinearity was assessed, with all variables having a variable inflation factor (VIF) < 2. We excluded three items from the analysis, as not enough respondents answered them: being physically hurt by the floods, having loved ones who were physically injured, and having lost a loved one during the floods. All other included items of the TESS are included in Appendix B To assess differences in MHC scores between men and women, an independent t-test was conducted. For all tests, we set the p-value at 0.05 and used statistical software SPSS 29.

For the qualitative analysis, a thematic, inductive approach examined the data until lead researcher NDM could not identify new codes. We combined these codes into categories to define themes. The percentages provided in this analysis are relative to the number of respondents who left open comments and not the total number of survey respondents. Figure 2 shows a summary of the methodology.

3. Results

3.1. Characteristics of the Study Sample

The study’s sample includes 114 persons (55.6%) who fully completed the questionnaire, out of 205 responses. We excluded responses when participants did not fill out enough questions to assess their psychosocial well-being or exposure level or if they were not personally affected by the floods. Of 114, 50 respondents shared open comments (43.9% of the total sample), of which 64 were available for qualitative analysis. Figure 2 shows the flow chart.

Of the respondents, 36.0% identified as male, 63.1% as female, and 0.9% chose not to disclose their gender. People from 19 distinct municipalities within the affected region responded, with a dominant representation of the Arrondissement of Verviers, which accounted for 73.7% (n = 84) of the total sample. Within Verviers, municipalities such as Trooz (24.6%) and Theux (20.2%) had particularly high response rates, reflecting the severity of the disaster’s impact in these localities. Other arrondissements, such as Liège (10.5%, n = 12) and Huy (5.3%, n = 6), were also represented, albeit to a lesser extent. Smaller proportions of participants came from the arrondissements of Marche-en-Famenne (2.6%), Dinant (2.6%), and Charleroi (0.9%), indicating more limited flood exposure or lower engagement from these areas.

Table 1. Demographic characteristics of the respondents.

Variable	Categories	Frequency	Percentage
Age	18–34	11	9.6
	35–44	31	27.2
	45–54	23	20.2
	55–64	28	24.6
	65 or older	21	18.4
	Prefer not to answer	0	0
Self-reported gender	Male	41	36.0
	Female	72	63.1
	Prefer not to answer	1	0.9
Socio-economic status	Very difficult–difficult to make ends meet	25	21.9
	Rather difficult to make ends meet	24	21.1
	Rather easy to make ends meet	32	28.1
	Easy–effortless to make ends meet	22	19.3
	Prefer not to answer	11	9.6
Education	≤Lower secondary education	18	15.8
	Higher secondary education	31	27.2
	Higher education	64	56.1
	Prefer not to answer	1	0.9
Civil status	In a relationship	76	66.7
	Single	37	32.5
	Prefer not to answer	1	0.9
Arrondissement of residence *	Verviers (Liège Province)	84	73.7%
	Liège (Liège Province)	12	10.5%
	Huy (Liège Province)	6	5.3%
	Marche-en-Famenne (Luxembourg Province)	3	2.6%
	Dinant (Namur Province)	3	2.6%
	Charleroi (Hainaut Province)	1	0.9%
	Prefer not to say	2	1.8%
Reported distance to the water	≤200 m	80	70.2
	200–500 m	16	14.0
	500–1000 m	4	3.5
	1000–1500 m	4	3.5
	1500–2000 m	0	0
	≥2000 m	3	2.6
	I do not know	4	3.5

* In Belgium, an arrondissement is an administrative district that sits between the provincial and municipal levels of government. Belgium is divided into 10 provinces, and each province is further divided into arrondissements. These arrondissements serve both judicial and administrative purposes.

shows a full summary of the demographic characteristics.

For reference, the population in the area was the following. In 2021, the population of Trooz was 8312 [28], while by 2024 it had declined slightly to 8058 [28]. At the scale of the entire Vesdre Valley, the population stood at 79,905 inhabitants in 2021, increasing modestly to 80,304 in 2024 [29]. According to 2024 Walstat data, the average age of the population in the Vesdre Valley is 40.0 years, with an average household size of 2.26 persons [30]. Household structures are diverse: 13.4% are married couples without children [30]. In terms of gender distribution, the Vesdre Police Zone, which covers this region, has 39,270 men and 41,034 women, with 79,029 people living in private households [30]. This reflects broader demographic trends in Wallonia, where 20.0% of the population is aged 65 or older, and 34% of households consist of a single person [31]. Additionally, 15% of Wallonia’s population is foreign-born, illustrating the demographic diversity that contributes to the social complexity and disaster vulnerability of the Vesdre Valley [31]. Understanding these demographic dynamics is essential for interpreting the impacts of the 2021 floods and planning future resilience strategies for this region.

3.2. Medium-Term Psychosocial Impacts Post-Flood

The sample’s PSC score distribution (Figure 3) shows that 38 respondents (33.3%) reported no changes in their psychosocial well-being, 33 (28.9%) indicated slight deterioration, 29 (25.4%) reported a strong deterioration, seven respondents (6.1%) reported a slight improvement, and seven respondents (6.1%) indicated a strong improvement. The mean MHC score also increased as the PSC score increased (Figure 4). The mean total MHC score in the function of current psychosocial well-being was compared to before the floods. To collect data on the respondents’ psychosocial well-being, we used the validated Mental Health Component (MHC) of the RAND-36 questionnaire. The MHC consists of the following blocks: mental health, vitality, social functioning, and limited role functioning due to emotional concerns [24,25]. The Cronbach’s alpha is 0.53. With these blocks, this questionnaire allows a thorough scope of the current psychosocial well-being of the respondents. For this research, we used a validated French translation [25,26]. We scored the responses to the RAND-36 following the steps defined in the literature, resulting in an MHC score out of 100 for each participant, with a higher score corresponding to higher well-being [24,27]. In addition to the MHC scale, the survey included a 5-point scale to assess whether the respondents’ psychosocial well-being has improved, deteriorated, or remained the same as before the floods. We used the following scoring: 1—strong deterioration, 2—slight deterioration, 3—no change, 4—slight improvement, and 5—strong improvement, meaning that a higher score corresponds with higher well-being. We refer to this variable as the psychosocial change (PSC) score to indicate the floods’ psychosocial impact directly.

In the qualitative data, 27 respondents (54.0% out of 50 respondents) mentioned a form of psychosocial impact after the floods: (1) fear of extreme rain and precipitation as well as fear of recurring events (34.0%); (2) being psychosocially hurt by the floods (20%), expressing it as being “psychosocially wounded”, “psychosocially hurt”, or “traumatized”, or by mentioning psychosocial issues they experienced; and (3) feeling destroyed or ruined in terms of psychosocial, physical, and financial health (6%).

In addition, seven respondents (14%) discussed attachment to their homes and nostalgia for what was lost, of which six (12%) discussed feelings of home related to the desire or necessity to move and three (6%) the drastic changes between life pre- and post-flood.

Table 2. Examples of reflections from stakeholders and community members about the psychosocial impact after the floods.

Demographics	Psychosocial Scores	Quote
Man 18–34 Relatively difficult to make ends meet	PSC: 1/5 MHC: 25/100	“I live with fears that did not exist before. I have fears of rain, thunderstorms, losing everything again to the point of not wanting to buy non-essential items anymore, etc.”
Woman 65–74 Relatively easy to make ends meet	PSC: 2/5 MHC: 79/100	“I experience constant stress during heavy rainfall, and I don’t think I could bear being flooded again.”
- 35–44 Easy to effortless to make ends meet	PSC:4/5 MHC: 55/100	“I went through acute depression during the six months following the floods. Then, I experienced phases of melancholy for about one year.”
Woman 55–64 Difficult to very difficult to make ends meet	PSC: 1/5 MHC: 6/100	“These floods have utterly destroyed our physical, mental, and financial health. We will be marked for life in all these areas. We have nightmares with every drop of rain.”
Woman 45–54 -	PSC: 1/5 MHC:8/100	“As time passes, everything seems more complicated… the more time passes, the more we realize that we will never be the same as before…”

highlights some quotations.

3.3. Demographic Variables Explaining Impact Variability

The ANOVA test for SES differences in PSC was highly significant (p = 0.003). The mean score rose gradually between the groups having a very hard to hard time making ends meet (1.80) versus those having some difficulty (2.25), those having a relatively easy time (2.88), and those having an easy to effortless time (2.64).

Table 3. Overview of multiple-comparisons ANOVA SES-PSC score used in this assessment.

Socio-Economic Status	Economic Status	Mean Difference	Significance	95% Confidence Interval
				Lower Bound	Upper Bound
A. Difficult–very difficult	Rather difficult	−0.450	0.140	−1.05	0.15
	Rather easy	−1.075	<0.001	−1.63	−0.52
	Easy–effortless	−0.836	0.008	−1.45	−0.22
	Prefer not to answer	−0.291	0.449	−1.05	0.47
B. Rather difficult	Difficult–very difficult	0.450	0.140	−0.15	1.05
	Rather easy	−0.625	0.031	−1.19	−0.06
	Easy–effortless	−0.386	0.219	−1.01	0.23
	Prefer not to answer	0.159	0.680	−0.60	0.92
C. Rather easy	Difficult–very difficult	1.075	<0.001	0.52	1.63
	Rather difficult	0.625	0.031	0.06	1.19
	Easy–effortless	0.239	0.417	−0.34	0.82
	Prefer not to answer	0.784	0.036	0.05	1.52
D. Easy–effortless	Difficult–very difficult	0.836	0.008	0.22	1.45
	Rather difficult	0.386	0.219	−0.23	1.01
	Rather easy	−0.239	0.417	−0.82	0.34
	Prefer not to answer	0.545	0.165	−0.23	1.32

below lists pairwise comparisons for PSC scores between SES groups, with significant differences in PSC scores between groups A and C, A and D, and B and C, without controlling for confounding effects such as the differences in PSC for gender or age per SES group.

We also assessed differences in MHC scores for SES, showing a significant result between groups A and C, A and D, and B and D.

Attained level of education had no significant effect on PSC (p-value: 0.292). The result was significant for the MHC score (p = 0.013) for those with higher education versus higher secondary education.

Comparing the average PSC and MHC scores for men versus women, the results show 2.46 and 2.32 versus 51.5 and 44.5, respectively, with no significant difference (p-value MHC: 0.07, p-value PSC: 0.316).

The ANOVA test reflected whether variables such as age, civil status, and proximity to water influenced participants’ PSC and MHC scores (

Table 4. Structured representation of one-way ANOVA analyses assessing differences in mean PSC score for age, civil status, municipality, insurance status, and residence’s distance to the water.

Variable	Categories	Mean PSC Score	p-Value (PSC)
Age	18–34	2.55	0.829
	35–44	2.19
	45–54	2.52
	55–64	2.39
	≥65	2.43
Civil status	Various	-	0.940
Distance to water	Various	-	0.430
Municipality	-	Sample size too low for analysis	-
Insurance status	-	Not tested (all but one respondent was insured)	-

). For age, five groups and the average PSC scores for these groups reflected the differences between the groups, indicating that age did not meaningfully impact PSC scores. Similarly, no significant differences were found in MHC scores based on age (p-value: 0.257).

We also examined the effect of civil status (e.g., single or married) on both PSC and MHC scores and found no significant impact (p-values: 0.940 and 0.405, respectively). Additionally, participants’ distance from the water did not significantly influence PSC (p-value: 0.430) or MHC scores (p-value: 0.887). Due to the few respondents from each municipality, we could not conclusively analyze differences between municipalities. Furthermore, we did not assess the effect of insurance status since nearly all respondents were insured except for one individual. Overall, the findings indicate that age, civil status, and proximity to water did not singly or explicitly affect participants’ psychosocial or mental health outcomes in this study; however, they may have acted in a combined manner in certain specific cases.

3.4. The Impact of Level of Exposure on the Experienced Psychosocial Impact

Multiple linear regression assessing the impact of factors of exposure (e.g., having to relocate) on PSC score was highly significant (p-value: 0.002), but the adjusted R² (0.157) showed a low explained variance, with only “having financial difficulties due to the floods” significantly contributing to the model (p = 0.002). Similarly, exposure significantly predicted MHC score (p-value < 0.001), but R² = 0.209, with “having suffered financial difficulties due to the floods” and “having had to relocate after the floods” explaining the model.

Qualitative data show that seven respondents (14.0%) mentioned the infrastructural damage (e.g., to the house or garden, heating or electricity), and six respondents (12.0%) mentioned the burden of ongoing repairs: “It has been 877 days since we were stuck in construction, and we are still there” (65+-year-old man, PSC: 1/5 and MHC: 14/100).

Respondents also related to the broader (post-)disaster context in the qualitative data. Dissatisfaction with emergency response and measures taken during and post-floods was mentioned by 11 respondents (22.0%). Of those, seven respondents explicitly mentioned the lack of support from authorities and the absence of strong anti-flood policies (e.g., prohibition of new construction in flood-prone areas), three respondents discussed the cumbersome interactions with insurance companies, and one respondent mentioned the lack of psychosocial support. Furthermore, five respondents (10%) felt “abandoned” during the response and reconstruction measures: “We will never forget. We have been abandoned by everyone (government, insurance companies, etc.). It’s a terrible shock” (55–64-year-old woman, PSC: 1/5 and MHC: 5/100). Lastly, nine respondents (18.0%) discussed additional stressors exacerbating psychosocial impacts, e.g., the loss of pets or farm animals, the stress of evacuating with small children, and going through evacuation when in advanced pregnancy.

4. Discussion

To the authors’ knowledge, this is the first study assessing the medium-term psychosocial impacts of the 2021 floods. We found that people affected by the 2021 floods in Southern Belgium continue to experience deteriorated psychosocial well-being >2 years after exposure, with >50% of participants reporting a decline. On average, those with a lower PSC score also had a lower MHC score indicating that those with the highest psychosocial impact post-floods also experience lower psychosocial well-being. The impact of the floods could be strong enough to affect general psychosocial well-being, but possibly, those reporting the highest psychosocial impact already had lower baseline well-being.

4.1. Key Findings

SES was the only demographic factor significantly influencing psychosocial well-being, with lower SES associated with more significant deterioration, which was also evident in the qualitative analysis. About 66.7% of respondents discussing the psychosocial impact of the floods reported difficulty making ends meet. Those with lower SES typically lived in higher flood-risk areas, meaning the most exposed often had the lowest SES [3]. This aligns with studies showing that SES affects the psychosocial impacts of floods [32,33] and the work by Lowe et al. (2013), who explain that SES influences a person’s ability to protect themselves and their property and their capacity to rebuild. This repair burden was also discussed in 12% of comments in our study. A decline in SES due to floods can also cause additional stress [33,34]. The sample size in our study was too small to perform a multivariate analysis, which could have controlled for other variables that may affect the impact of SES.

Current research predominantly evaluates the impact of flooding on psychosocial well-being in the immediate aftermath of exposure. Studies show that shortly after experiencing a flood, affected populations frequently report psychiatric symptoms such as anxiety, depression, despair, indignation, somatic dysfunctions, and nightmares [35,36]. According to the National Collaborating Centre for Environmental Health (NCCEH), individuals who decide to remain in flood-prone areas despite evacuation warnings may experience stigma and social isolation from their communities, compounding their emotional distress [37]. The financial burden associated with rebuilding after flood damage, particularly among those with limited financial or social support, can further strain relationships and intensify psychosocial difficulties [33,38].

While floods clearly disrupt psychosocial well-being in the short term, emerging research highlights the importance of considering long-term effects, which often do not surface immediately. Psychological distress can manifest months or even years after the event, fluctuating over time or, in some cases, becoming chronic [35,36]. In economically disadvantaged communities, the prevalence of psychological disorders can even continue to increase over the long term, as structural vulnerabilities compound the mental health burden [39]. This underscores the need for longitudinal research that extends beyond the immediate post-disaster phase to capture the evolving psychosocial consequences of flooding, especially for vulnerable populations living in high-risk flood zones [37].

Findings from this study align with these broader patterns but also highlight some context-specific dynamics within the Vesdre Valley. Consistent with earlier UK-based research [40], our data confirm that financial difficulties caused by the flood had a significant impact on psychosocial well-being. However, as Fernandez et al. (2015) noted, inconsistent definitions of flood exposure in existing research limit the comparability of findings and the precision of exposure–outcome relationships [20]. Our qualitative data reinforce this, with 14.0% of respondents’ comments describing infrastructural damage and 12.0% mentioning ongoing repair burdens, highlighting the importance of refining flood exposure measurement tools to include physical damage and prolonged reconstruction stress.

Additionally, existing research has shown that socioeconomic status (SES) moderates the relationship between flood exposure and psychosocial impact, with lower SES groups experiencing greater distress [20]. While our sample size was too small to statistically test this interaction, qualitative comments reflect a sense of financial precarity among some respondents, who expressed feeling overwhelmed by repair costs and administrative complexity. This aligns with evidence from post-flood studies that identified cumbersome insurance processes and insufficient financial support as key stressors amplifying distress [20,41,42]. In our study, 6.0% of participants explicitly criticized insurance responses, while 22.0% expressed dissatisfaction with broader government recovery support, mirroring findings from UK and Canadian research [20,33].

Finally, our study confirms that psychological distress can persist well beyond the initial disaster phase, influenced by the pace of recovery and perceived support from authorities. This resonates with the growing recognition that long-term mental health consequences are not just a product of the initial disaster experience but also of how recovery processes unfold in the years that follow [38]. Especially in flood-affected areas with economic vulnerabilities, delayed recovery and perceived neglect by authorities can fuel chronic distress and erosion of community trust [20,33,38]. This highlights the critical need for integrated disaster response strategies that address both immediate and long-term psychosocial recovery, while explicitly acknowledging social inequalities that shape these processes [20,33,38].

4.2. Policy Implications

First, while the health effects of climate change are gaining attention, most studies focus on short-term impacts [19]. Research must include long-term effects, especially on vulnerable groups and understudied regions. To advance flood impact research, one may need a standardized regional definition of exposure to clarify its effects on psychosocial outcomes and enable cross-population comparisons.

Second, public health policies should integrate better climate-sensitive approaches, including psychosocial support during and after climate-related disasters. The 2021 floods revealed the need for better coordination between emergency and psychosocial services. Psychosocial support should extend beyond the short term, as impacts persist for over two years post-disaster. The Australian National Disaster Mental Health and Wellbeing Framework (2023) offers a model of a multi-level psychosocial support plan and strategies to strengthen community resilience [43]. The Netherlands also developed an intervention plan targeting vulnerable groups and promoting self-reliance [44].

Third, effective insurance schemes for rapid coverage of climate-related costs are crucial to bolster the resilience of socio-economically disadvantaged groups [43,45]. Structural measures should address the social housing gradient in flood-prone areas, a key step to reducing post-flood psychosocial impacts on low-SES communities.

Finally, public awareness of climate risks should improve. Many perceive these risks as distant, leading to under-preparation [46,47,48,49]. Adaptive risk cultures, which foster investment in mitigation, prevention, and preparedness, can reduce psychosocial impacts by shortening exposure and enhancing coping strategies. Government officials could promote public awareness through information sessions, drills, and education. Communication should leverage official channels and wide-reaching citizen networks [49,50].

4.3. Strengths and Limitations of the Study

Our study also has a number of strengths and limitations. First, the analysis incorporates a wide range of demographic characteristics, including age, gender, income, education, and those who preferred not to answer. However, segmenting our sample of 114 participants across all possible combinations of these variables resulted in small subgroup sizes, as shown in Table 2. For example, when considering the intersection of age brackets, gender identity, income levels, and educational attainment, the number of respondents fitting each unique demographic profile became very limited. As a result, the statistical power for making meaningful comparisons between these highly specific groups was significantly reduced. Therefore, while these demographic variables provide valuable contextual information about our respondents, the insights derived from such small subgroup comparisons must be interpreted with caution. Given this limited statistical power, observed differences between groups may not reach statistical significance and may not be generalizable to a broader population. In future research, we aim to mitigate this limitation by either expanding the sample size or focusing on a more targeted set of demographic variables, ensuring sufficient statistical power for subgroup analysis. Our goal is to strike a careful balance between capturing demographic diversity and maintaining the statistical rigor necessary to draw reliable conclusions.

Although the online format of the survey presents certain limitations—including potential overrepresentation of particular groups, such as younger adults or those with higher education—the overall findings remain representative. This is because they align with existing research and draw upon standardized, validated methods, ensuring that the results are transferable. Importantly, this study is the first to quantitatively assess the medium-term psychosocial impact of the 2021 floods in Southern Belgium and Western Europe, contributing to the limited evidence available on this topic. To enhance reliability and comparability, the study employed validated tools, including the SF-36 to assess well-being and the TESS to measure disaster exposure. This approach enabled us to capture insights from a broad population across 20 municipalities, providing valuable evidence on the psychosocial consequences of the flooding.

Recognizing the value of qualitative depth, we acknowledge that semi-structured interviews could have offered richer insights into residents’ concerns and experiences, allowing for a deeper understanding of their perspectives. To address this, we incorporated notes from semi-structured interviews conducted as part of a related publication, complementing our quantitative findings and providing a more comprehensive discussion that captures the complexities of residents’ experiences following the floods.

At the same time, it is essential to acknowledge certain methodological limitations inherent in this study. One key limitation is the absence of a second independent researcher for the qualitative analysis. While involving an additional researcher could have strengthened reliability and validity by introducing alternative interpretations, this was not feasible due to resource constraints. Instead, the primary researcher, Ms. Charlotte Scheerens, who possesses in-depth knowledge of the local context and has actively engaged with the community over several years (2021–2023), conducted the analysis. Her familiarity with the affected communities was instrumental in interpreting the nuanced, place-based narratives central to this study. Overall, consistency in narrative interpretation was maintained, aligning with the exploratory nature of our research design, which prioritizes depth over breadth. To mitigate potential biases, we employed data triangulation, drawing on multiple data sources, including policy documents, interviews, and field observations, to enhance the robustness of our findings.

Finally, while the online survey format may have introduced selection bias—for instance, underrepresenting children and amplifying responses from more educated participants—and while the modest sample size limited the scope for deeper multivariate analysis, the study nonetheless offers critical insights. In particular, it highlights the absence of data on children, alongside potential challenges in measuring disaster exposure. Despite these limitations, the study’s findings provide valuable guidance for enhancing disaster response and recovery efforts, particularly concerning psychosocial support. Ultimately, these insights underscore the need for continued research to further refine and strengthen disaster preparedness and community resilience in the face of future climate-driven events.

5. Conclusions

Our study shows that adverse psychosocial effects persist more than two years post-event, illustrating the necessity for continued support mechanisms. We noted a significantly higher psychosocial impact on individuals with lower socio-economic status. Other demographic factors, traditionally considered vulnerability indicators, did not considerably influence the psychosocial impact in this sample. Higher exposure to floods also contributed significantly to the psychosocial impact experienced after the floods. Future disaster response and preparedness efforts should address this apparent vulnerability.

Following these findings, we formulated four key suggestions for improved resilience and preparedness for climate-exacerbated disasters and their psychosocial impact. First, we emphasize the need for further research to clarify the pathways linking climate-exacerbated disasters to psychosocial well-being. Second, we reflect on the need for climate adaptation strategies and public health policies that incorporate emergency and medium-term psychosocial support measures, focusing on the most vulnerable groups. Third, we emphasize the importance of financial support and insurance coverage for flood-related damages. Fourth, we underline the need to promote community awareness of disaster risks and to enhance community knowledge of response strategies to improve resilience and self-reliance. This study is a vital call to action for integrating psychosocial considerations into climate adaptation and disaster response frameworks. Doing so can ensure a more holistic and practical approach to safeguarding the well-being of populations vulnerable to extreme weather events’ increasing frequency and severity.