Professional Documents
Culture Documents
Public Health
Original Research
Article history: Objectives: We aimed to study the risk of developing post-traumatic stress disorder (PTSD)
Received 26 July 2016 symptoms in people who resided in an affected area by an extremely severe flood, and
Received in revised form sociodemographic risk factors associated with this condition.
7 December 2016 Study design: A geographic information system (GIS) was used to distribute the rainfall data.
Accepted 13 December 2016 A case-control study was developed to study the relationship between PTSD and socio-
Available online 12 January 2017 demographic risk factors.
Methods: To delineate the areas affected by the flood and the intensity of this rainfall in
Keywords: comparison with historical hydrological data, we employed geographical information
Flood systems (GIS). Then, we recruited a representative sample of the affected population and
Flooding another population sample that lived at the time of this disaster in adjacent geographical
Natural disaster areas that were not affected. Both groups were randomly selected in primary care prac-
Post-traumatic stress disorder tices, from December 1st 2012 to January 31st 2013. All participants, 70 from the affected
Public health areas and 91 from the non-affected, filled a sociodemographic questionnaire and the
trauma questionnaire (TQ) to identify and rate PTSD symptoms.
Results: Our GIS analysis confirmed that the amount of precipitation in 2012 in the areas
affected by the flood was exceptionally high compared with historical average rainfall data
(461l per square metre vs 265). Individuals who resided in the affected areas at the time of the
flood were at much higher risk of developing PTSD symptoms (OR: 8.18; 95% CI: 3.99e17.59)
Abbreviations: DTM, digital terrain model; GIS, geographical information systems; IDW, inverse distance weighed; PTSD, post-trau-
matic stress disorder; TQ, trauma questionnaire.
* Corresponding author. Area Gestion Sanitaria Norte Almeria, Avda. Dra Ana Parra SN, Huercal-Overa, Almeria, 04618, Spain. Fax: 34
950451591.
E-mail address: andresfontalba@gmail.com (A. Fontalba-Navas).
i
Joint senior authors.
http://dx.doi.org/10.1016/j.puhe.2016.12.015
0033-3506/ 2016 The Author(s). Published by Elsevier Ltd on behalf of The Royal Society for Public Health. This is an open access article
under the CC BY-NC-ND license (http://creativecommons.org/licenses/by-nc-nd/4.0/).
p u b l i c h e a l t h 1 4 4 ( 2 0 1 7 ) 9 6 e1 0 2 97
than those living in adjacent, non-affected localities. Among the sociodemographic variables
included in this study, only material and financial losses were strongly associated with the
onset of PTSD (P < 0.001). Physical risk during this life-threatening catastrophe also indicated
a positive correlation with later development of PTSD symptoms; however, it did not reach
statistical significance (P 0.06).
Conclusions: Populations affected by severe floods may suffer an increase of PTSD symp-
toms in the following months. This finding, along with the importance of material losses as
a predictor for such disorder, may help develop effective plans to minimize the negative
impact of these natural disasters on public health.
2016 The Author(s). Published by Elsevier Ltd on behalf of The Royal Society for Public
Health. This is an open access article under the CC BY-NC-ND license
(http://creativecommons.org/licenses/by-nc-nd/4.0/).
database, including information from 35 weather stations events, including natural disasters. It also collects information
distributed across the study area. We collected data for a regarding the age at which the individual was exposed to any
period of 78 years (1935e2012). We completed missing infor- of these events and the length of exposure. The second section
mation by carrying out correlation analyses with nearby explores the specific traumatic events that the patient de-
stations. scribes as most severe/worrying and confirms whether he/she
The rainfall data analysis allowed us to create a spatial meets DSM-IV criterion A for PTSD.13 Physical risk and asso-
map of precipitations in the study area. A geographic infor- ciated stressors are identified in this section. The final section
mation system (GIS) was used to distribute the rainfall data. contains 18 items, a list of symptoms, with dichotomic re-
With the data obtained from each weather station, we applied sponses (YES/NO), to explore the presence of DSM-IV diag-
the following stepped methodology: nostic criteria BeD for PTSD. It also provides a score that rates
the severity of the current presentation.
1. Interpolation method of the rainfall data. The Inverse Material damages and losses caused by the flood were also
Distance Weighted (IDW) method was used to interpolate recorded, but only if the estimated costs associated with res-
annual tendencies of the climatic variables analyzed at torations or replacements had been fixed by a judiciary au-
each station. This deterministic interpolation technique is thority and a damage evaluation expert opinion.
one of the most frequently used for the creation of spatial This study was approved by the Ethical Committee of
interpolation maps.9 Clinical Investigation Almeria Centro (Spain).
2. Verification of basic statistics for the data.
3. Procurement of a digital terrain model (DTM) using rainfall Sample size
as spot height.
4. Contour lines (isolines) were made at certain equidistance. To detect differences in the incidence of PTSD symptoms be-
5. Data interpretation using the isolines obtained with the tween the population affected by the flood and those non-
GIS.10 affected living in adjacent municipalities, for a power of 90%
with a significance level of 5%, using a bilateral Chi-squared
test for two independent variables and assuming a propor-
Study population and data collection tion of PTSD in the affected group of 22% and 1.95% in the non-
affected, we needed to include 53 participants in each group.
We selected a representative sample of the directly (and
severely) affected population in the municipalities of Vera, Statistical analysis
Cuevas del Almanzora and Pulpi in Almeria, which covered a
population of 37,154 in 2012 (Spanish National Institute of We compared the sociodemographic characteristics between
Statistic; http://www.ine.es). Another population sample that the population affected by the flood and the non-affected. We
lived at the time of this disaster in adjacent (less than 30 kms employed t-test to compare the age of both groups and Chi-
from the affected areas) municipalities (Zurgena, Macael and square test for the sex distribution. For educational levels
Albox) that were not affected by the flood was also included in and occupational status, we used ANOVA. To determine the
this study. These municipalities covered a total population of difference in the risk of developing PTSD symptoms between
20,835 in 2012. both groups (odds ratio; OR), we employed logistic regression
All participants were recruited by four General Practitioners adjusted for sex, age, educational level and occupational
(GPs), two of them working in primary care practices covering status.
the municipalities affected by the flood and the other two in the We used multivariable linear regression to evaluate factors
non-affected. In Spain, primary care usually is the first point of associated with the onset of PTSD symptoms, according to the
contact for people with mental health problems. Everybody is TQ. We created a single model with the following independent
entitled to receive public health care; thus primary care prac- variables: sex, age, educational level, occupation and
tices are ideal places to detect people with emerging psychiatric measurable impacts (physical risk and financial/material
disorders, such as PTSD.11 We employed a simple systematic, looses). The linear regression model was adjusted for sex and
pragmatic sampling methodology, where GPs offered partici- age regardless of their statistical significance, since PTSD is
pation in the study, one to every five consecutive patients, aged more common in women14 and the elderly.15
18 or over, seen in their clinics for any reason, at least 2 months P-values 0.05 were considered as statistically significant.
after the disaster; between December 1st 2012 and January 31st All analyses were performed using version 3.2.0 of R.16
2013. We checked, with GIS, whether those patients who
expressed an interest to take part resided in areas affected by
the flood on September 28th 2012. Results
After obtaining written consent, each participant filled a
basic sociodemographic questionnaire, including sex, age, In 2012, the yearly amount of precipitation registered in the
educational level and occupational status, and a validated affected area significantly exceeded the average yearly pre-
Spanish version of a trauma questionnaire (TQ)12,13 to identify cipitation for the previous 78 years (1935e2012). For example,
and rate PTSD symptoms. the average yearly precipitation registered at the municipality
The TQ is a self-report questionnaire that comprised three of Huercal Overa in 2012 was 461l per square metre and the
sections. It begins with a screening that determines whether historical average was 312l. In Vera, the local weather station
the patient suffered any of the 17 different types of traumatic registered 454l, whilst the historical average was 295. Thus, for
p u b l i c h e a l t h 1 4 4 ( 2 0 1 7 ) 9 6 e1 0 2 99
these areas, the amount of precipitation in 2012 was excep- the onset of PTSD symptoms and the flood, which was rigor-
tionally high. This difference is clearly shown in Figs. 1 and 2, ously analyzed with hydrological and statistical methods.
which display a digital terrain model (DTM) with the average A previous meta-analysis suggested that trauma severity,
yearly precipitation in each map. lack of social support and additional life stressors were strong
A total of 70 individuals affected by the flood and 91 from predictors of PTSD. However, these results were limited by the
the adjacent areas participated in this study. Using the sam- variety of measures used to determine trauma severity, both
pling methodology outlined above, GPs only had to approach between and within different types of traumas.18 Focusing on
approximately 100 individuals in the affected areas and 150 in traumatic events, such as natural disasters, that allow the use
the non-affected to recruit these samples between December of more accurate and objective measures of severity can
1st 2012 and January 31st 2013. The characteristics of both reduce this limitation. For example, GIS accurately pinpointed
groups are summarized in Table 1. There were no statistical the geographical radius affected by this catastrophe and its
differences between the two samples with regard to sex, age, intensity.
educational level or occupational status. Climate change is affecting the spatio-temporal variations
The likelihood to suffer PTSD symptoms (OR) was 8.18 of extreme precipitation. It is expected that variations in fre-
times higher (95% CI: 3.99e17.59) in the population affected by quency and intensity of 24-h extreme precipitation will favour
the flood. the increase of torrential rain processes and dynamic
Among the variables included in this study, only material geomorphic hazards which may increase human and material
losses were strongly associated with the onset of PTSD (see losses. Therefore, the consequences of a catastrophic flood on
Table 2). Interestingly, physical risk was also positively asso- a population are mainly related to the intensity and impact of
ciated with the later development of PTSD symptoms, the disaster. However, sociodemographic factors may also
although the association was only borderline-significant. play an important part. Notably, the main sociodemographic
variable that predicted the onset of PTSD symptoms in our
study was the material and financial losses associated with
Discussion the disaster. Interestingly, this association was stronger than
the perception of serious physical risk at the time of this life-
Extraordinary torrential rain that results in a severe flood may threatening event. Our finding is in line with previous studies
cause severe distress in individuals affected, not only during that identified financial difficulties after domestic floods as a
the event, but also in the aftermath.17 Our study shows that risk factor to develop severe mental disorders.19,20
people affected by floods may subsequently suffer symptoms One of the limitations of the study is its cross-sectional
of PTSD. Since PTSD may be a prevalent disorder in the gen- nature. We recruited participants and collected data be-
eral population, our comparison of a population affected by a tween 2 and 4 months after the catastrophe. We did not follow
severe flood in Southern Spain with another, non-affected, them up and could have missed some cases that developed
living in adjacent areas reinforced the association between PTSD symptoms after a longer latent period. This would have
Fig. 1 e Raster image representing average yearly precipitation in the study area from 1935 to 2012.
100 p u b l i c h e a l t h 1 4 4 ( 2 0 1 7 ) 9 6 e1 0 2
Fig. 2 e Raster image representing average yearly precipitation in the study area for 2012.
increased the incidence of this condition rather than sample during the study period. Most of them were recruited
decreasing it. Also, the study only included people recruited in consecutively, following the sampling methodology outlined
primary care practices, and not all the patients invited finally above. Furthermore, everybody in Spain has access to the
agreed to take part in the study, which might affect the gen- National Health System, which provides public healthcare for
eralisability of our results. However, as stated before, GPs did Spanish citizens, nationals of European Union (EU) Member
not have to invite a large number of people to recruit our final States and non-EU Member States, and other foreign na-
tionals, including undocumented immigrants within the
Spanish national territory.21 Thus, 100% of the population is
entitled to be treated in primary care practices.22 Accordingly,
Table 1 e Sociodemographic characteristics of the groups
we believe that our sample was representative of the people
affected and not affected by the flood.
Variable Affected Non- P-value
affected
N 70 91 Table 2 e Multivariable linear regression analysis to
Age (mean, SD) 51.58 (15,42) 51.74 (17,45) 0.95a determine predictors of PTSD symptoms according to the
Sex (% female) 64.03 62.60 0.82b TQ.
Education level (%) 0.87c
Variable Beta Standard P-
No education 1.4 1.1
error value
Primary school 24.3 22.0
High school 45.7 47.3 Intercept 0.28 2.38 0.90
University 28.6 29.7 Age 0.02 0.04 0.62
Occupational status (%) 0.64c Sex (female) 0.70 0.76 0.36
Employed 35.7 29.7 Occupation (housewife)a 0.14 1.32 0.91
Self-employed 8.6 11.0 Occupation (unemployed)a 0.11 1.14 0.92
Housewife 14.3 9.9 Occupation (retired)a 1.51 1.39 0.27
Unemployed 21.4 20.9 Occupation (self- 1.55 1.30 0.23
Retired 20.0 28.6 employment)a
TQ score (mean, SD) 5.39 (6.02) 1.59 (4.13) <0.01a Education (school)b 0.73 1.10 0.12
Physical risk during 12.90 2.20 <0.01b Education (high school)b 2.13 1.87 0.25
flood (%) Education (university)b 0.09 0.95 0.33
Material losses during 47.10 3.30 <0.01b Physical risk (yes) 2.85 1.47 0.06
flood (%) Financial losses (yes) 6.90 0.95 <0.01
a
t-test. Multiple R-squared: 0.3919; adjusted R-squared: 0.3457.
b a
Chi-square test. Reference: employed.
c b
ANOVA. Reference: no studies.
p u b l i c h e a l t h 1 4 4 ( 2 0 1 7 ) 9 6 e1 0 2 101
15. Moye J, Rouse SJ. Posttraumatic stress in older adults: when 20. Paranjothy S, Gallacher J, Amlo ^ t R, Page L, Baxter T, Wight J,
medical diagnoses or treatments cause traumatic stress. Clin et al. Psychosocial impact of the summer 2007 floods in
Geriatr Med 2014;30(3):577e89. England. BMC Public Health 2011;11:145. http://dx.doi.org/
16. The R Core Team. R: a language and environment for statistical 10.1186/1471-2458-11-145.
computing. Vienna: R Foundation for Statistical Computing; 21. Borkan J, Eaton C, Novillo-Ortiz D, Rivero P, Jadad A. Renewing
2014. primary care: lessons learned form the Spanish Health Care
17. Du W, FitzGerald GJ, Clark M, Hou XY. Health impacts of System. Health Aff 2010;29(8):1432e41.
floods. Prehosp Disaster Med 2010;25(3):265e7. 22. Ministerio de Sanidad. Servicios Sociales e Igualdad, 2016.
18. Brewin C, Andrews B, Valentine J. Meta-analysis of risk National Health Survey; 2016. http://pestadistico.
factors for posttraumatic stress disorder in trauma-exposed inteligenciadegestion.msssi.es/publicoSNS/Comun/
adults. J Consult Clin Psychol 2000;68:748e66. DefaultPublico.aspx.
19. Kishore V, Theall KP, Robinson W, Pichon J, Scribner R, 23. Lowe D, Ebi KL, Forsberg B. Factors increasing vulnerability to
Roberson E. Resource loss, coping, alcohol use, and health effects before, during and after floods. Int J Environ Res
posttraumatic stress symptoms among survivors of Publ Health 2013;10(12):7015e67.
Hurricane Katrina: a cross-sectional study. Amer J Disaster Med 24. Wood PJ, Hannah DM, Sadler JP. Hydroecology and ecohydrology:
2008;3(6):345e57. past, present and future. Chichester: John Wiley & Sons, Ltd; 2008.