The English National Cohort Study of Flooding & Health: psychological morbidity at three years of follow up

Flooding is the most common natural disaster worldwide and has been shown to have an adverse impact on both physical and psychological health [1]. In England, it is estimated that around 5.2 million properties are at risk of flooding [2]. Frequency and intensity of floods are anticipated to increase in the future due to population growth, urban development on flood plains, and climate change [3, 4].

In high-income countries such as the UK, the greatest burden of disease following flooding is adverse mental health outcomes [5]. In addition, displacement from homes can result in stress arising from dealing with household repairs and disruption to public services [6]. A number of factors have been found to increase vulnerability to experiencing psychological impacts following extreme weather, including older age, pre-existing medical conditions, inadequate insurance cover and social deprivation [7].

There is a paucity of studies quantifying the longer-term impacts of flooding on health, particularly beyond the first year post-flooding [8]. Following floods affecting England in 2013–2014, Public Health England (PHE) established the National Study of Flooding and Health (NSFH), to investigate the long-term impact of flooding and associated disruption on psychological health. The study aims to support preparedness and response activities to future flooding events.

The NSFH has previously identified a significant adverse impact on mental health, both at one and 2 years post-flooding, in those whose homes were flooded and whose lives were otherwise disrupted by flooding, compared with those unaffected [9, 10]. The NSFH also identified that adverse outcomes are associated with secondary stressors [11], such as insurance-related issues [12], and with displacement from home without warning [13]. In this study, we aim to assess mental health morbidity at 3 years post-flooding and the impact of persistent flood-related damage in the home. We also aim to assess the prevalence change over the three-year period, to identify possible predictors for psychological recovery.

Of the 1361 participants contacted at year three, 896 responded with a questionnaire, however 29 questionnaires were blank and were not included in the analysis (63.7% valid response rate). A further 48 exclusions were made, including 9 duplicates, 3 who reported a new episode of flooding and 36 who had a missing exposure status at year one. Of the 819 included in the final analysis, 119 (14.5%) were classified as unaffected, 421 (51.4%) as disrupted and 279 (34.0%) as flooded. Of those, 569 had completed the questionnaire in all 3 years, with 93 (16.3%) classified as unaffected, 289 (50.8%) as disrupted and 187 (32.9%) as flooded.

Overall, approximately 5.7% reported symptoms of probable depression, 8.1% of probable anxiety and 11.8% of probable PTSD, with the prevalence of all adverse probable mental health outcomes higher in the flooded group than unaffected (Table 1).

The adjusted odds ratio (aOR) of probable depression and PTSD were significantly higher in the flooded group compared with the unaffected group, with aOR 8.48 (95% CI 1.04–68.97) and aOR 7.74 (95% CI 2.24–26.79), respectively. The aOR of probable anxiety was elevated in the flooded group, compared with the unaffected group but not significantly (aOR 2.68, 95% CI 0.88–8.20). Participants who were disrupted by flooding had increased odds of PTSD (aOR 4.33, 95% CI 1.26–14.92), compared with the unaffected group; no other probable mental health outcomes in the disrupted group were statistically significant (Table 2).

Seventy-seven (9.4%) participants reported persistent damage to their home because of the original flooding episode. The most commonly reported issues were damp in liveable rooms (n = 40), visible mould in liveable rooms (n = 26), problems with damp or water in non-liveable rooms such as garage, cellar or basement (n = 12) and drains backing up and flooding (n = 10). After adjusting for potential confounders, those who reported persistent home damage at year three were statistically more likely to suffer from depression and PTSD, compared with those who did not report persistent damage, however anxiety was not significantly elevated in this group (Table 3).

We observed a prevalence change of probable mental health outcomes over three consecutive years post-flooding, for the 569 participants who completed the questionnaire in all 3 years (Fig. 1, Supplementary Table 1). In the flooded group, we observed a significant reduction in prevalence across all three probable mental health outcomes: depression (year one 20.8%, year two 11.2%, year three 7.8%, p = 0.0014), anxiety (year one 27.6%, year two 12.3%, year three 11.8%, p < 0.001) and PTSD (year one 33.2%, year two 24.9%, year three 17.1%, p = 0.001). The reduction was suggestive in the disrupted group for depression (year one 8.3%, year two 4.8%, year three 5.1%, p = 0.05) and in the unaffected group for PTSD (year one 5.6%, year two 0%, year three 1.9%, p = 0.045), but not for anxiety in either the disrupted or unaffected group. No significant predictors were identified for the reductions in prevalence of adverse mental health outcomes.

Few studies have focused on the long-term prevalence of mental health problems in those affected by flooding, with systematic mapping reviews by Zhong et al, 2018 and Fernandez et al, 2015 highlighting the lack of studies conducted on this topic two or more years post-flooding [1, 17]. Our paper assessed the prevalence of probable mental health outcomes 3 years after a flooding event.

We identified that the adverse impact of flooding on mental health persists for at least 3 years after the flooding event, with a higher prevalence of psychological morbidity (significantly for depression and PTSD) in flooded participants, compared with those unaffected. Many individuals reported persistent damage to their homes, which was a strong predictor for poorer mental health outcomes, compared with other people who were exposed (disrupted or flooded) but who did not report experiencing persistent damage issues.

Overall the data show a reduction in psychological morbidity over the 3 years in the flooded group, a suggestive decline in the disrupted group and no significant differences in the unaffected group. Unfortunately, we were unable to identify any predicators of this recovery in this cohort; previous studies has found factors such as availability of social support and personal coping style could influence recovery from PTSD post-flooding [18, 19]; however, these were conducted in China and more research is required on understanding the mechanisms for recovery in other contexts post-flooding.

We observed nearly half the prevalence of PTSD symptoms in the flooded group at year three post-flooding, compared with year one. This is in line with previous studies on PTSD related to natural disasters where one meta-analysis calculated a spontaneous remission rate of 60.0% [20], and may partly reflect the resolution of ongoing stressors that were helping to maintain distress in this group.

Within our study we observed 17.5% of flooded individuals with scores that indicated probable PTSD at 3 years after flooding, which is line with previous studies – 22% of individuals in South Korea at 18 months after flooding [21] and 8.6% of individuals 2.5 years after flooding in China [22] experienced probable PTSD. However, despite the decline in prevalence observed in people who have experienced flooding over the 3 years, there is still persistence of psychological morbidity, which may indicate a possible risk of chronic mental health problems if affected people do not receive suitable treatment.

This study has shown that the adverse impact of flooding on mental health persists for at least 3 years after the event, and that persistent damage to liveable rooms in the home is associated with more severe mental health outcomes. Work is needed to develop and evaluate interventions to increase resilience within populations at risk of flooding and to ensure prompt access to appropriate services following a flooding event.