behavior of crossing flood on foot, associated risk factors and estimating a predictive model
TRANSCRIPT
ORI GIN AL PA PER
Behavior of crossing flood on foot, associated risk factorsand estimating a predictive model
Hamidreza Shabanikiya • Hesam Seyedin • Hamid Haghani •
Abbasali Ebrahimian
Received: 14 August 2013 / Accepted: 1 March 2014� Springer Science+Business Media Dordrecht 2014
Abstract Crossing the flood on foot is one of the two major causes of flood-related death.
This study was aimed to determine risk factors associated with risky behavior of crossing
the flood on foot and modeling behavior of people when exposed to the flood. Data were
gathered by a questionnaire in Quchan, a city of Iran. People with the age of 18–35 years
old, those who do not take flood warnings seriously, individuals who do not have expe-
rience of exposure to flood and those who believe they have moderate to advanced level of
swimming skills were identified as high risk groups. Appropriate group training programs
can be set for them to reduce risky behavior of crossing the flood on foot.
Keywords Crossing on foot � Risk behavior � Flood � Iran
1 Introduction
Flood is the most common type of natural hazards and causes many disaster-related
casualties (Ciottone et al. 2006; Clements 2009). Iran is located in flood-prone areas, and in
recent decades, it has brought over four million homeless, eight thousand casualties and
many injured (EM-DAT). Moreover, frequency of flood occurrence is increasing, both in
the world and in Iran (Ardalan et al. 2009). The primary cause of flood-related morbidity
H. Shabanikiya � A. EbrahimianHealth in Emergencies and Disasters, School of Health Management and Information Sciences,Iran University of Medical Sciences, Tehran, Iran
H. Seyedin (&)Health Management and Economics Research Center, School of Health Management and InformationSciences, Iran University of Medical Sciences, Tehran, Irane-mail: [email protected]
H. HaghaniStatistics Department, School of Health Management and Information Sciences, Iran Universityof Medical Sciences, Tehran, Iran
123
Nat HazardsDOI 10.1007/s11069-014-1124-5
and mortality is drowning (Ciottone et al. 2006). Result of a study indicates that 93 %
flood-related death is due to drowning, of this percentage, 43 % is related to those who are
drowned in vehicles and 43 % are swept into the water while trying to cross the water on
foot (Coates 1999). Jonkman and Kelman (2005) reported that 70 % of flood deaths are
due to drowning, 33 % due to drowning in vehicles and 25 % due to drowning as a
pedestrian, This means that the main causes deaths are preventable two high risky
behaviors (driving into flood and crossing the flood water on foot). By examining and
analysis of these behaviors among populations located in the flood-prone areas, reaction of
people during flooding can be predicted. As a result, high risk groups can be identified and
proper educational material can be provided for the identified high risk groups in order to
reduce the incidence of these behaviors. The majority of studies that are conducted on
flood, as natural disaster, are about the rate and cause of mortalities due to flood (Fitz-
Gerald et al. 2010; Fillmore et al. 2011; Kellar and Schmidlin 2012) and diseases and
morbidities related to flood (Janerich et al. 1981; Tapsell and Tunstall 2008). Also, there
are some researches in the field of evaluation of preparedness for natural disasters, such as
earthquake and flood and assessment of behaviors and reactions of individuals against a
variety of these natural disasters (Tapsell et al. 2002; Kreibich et al. 2005; Coulston and
Deeny 2010; Molinari and Handmer 2011; Seyedin et al. 2011; Schad et al. 2012). Fur-
thermore, some studies have been carried out on flood risk assessment, the perceived risk
of flooding and the role of communication in this regard. In a study conducted by Sorensen
(Sorensen 2000), social factors were considered as potential risk factors influence the
behavior of people in disasters.
Few studies have been conducted on one of the two dangerous behaviors during a flood
i.e., driving into flood. Drobot et al. (2007) examined the association of some risk factors
and dangerous behavior of driving in flood. However, the researchers did not find any study
on dangerous behavior of crossing the flood on foot. This study is aimed to determine risk
factors associated with dangerous behavior of crossing the flood on foot and use of these
risk factors for modeling to predict the behavior of people when exposed to the flood.
2 Method
This study was performed in Quchan, which is located in northeastern of Iran, as one of the
areas prone to flooding (2012). A researcher-made questionnaire was used to collect the
data. The validity of the questionnaire was tested by experts in an expert panel.
A pilot study was performed to test the reliability of the questionnaire. The reliability of
questionnaire was proved by test–retest.
The questionnaire was arranged in two parts. The first part of the questionnaire was
devoted to assessing the dependent variable of study, i.e., risky behavior of crossing the
flood on foot. In this section, we want to determine whether the respondents say at the flood
situation will pass it on foot or not, they were given the following scenario:
Imagine you’re going to pass a constant flow of water, such as rivers or temporary
streams such as streams of water flowing in the streets, roads, valleys, and plains
after a heavy rainfall. What do you do in this situation?
Then they were asked to respond to six multiple-choice questions with a Likert scale
from ‘‘strongly agree’’ to ‘‘strongly disagree.’’ Internal reliability of these six questions was
measured by Cronbach’s alpha test, and with regard to alpha value obtained that was
Nat Hazards
123
greater than 7, internal validity of this part of questionnaire was proved. The first and last
questions of this section are listed below:
Question 1: I will cross the stream if its height is 25 cm (Approximately four fingers
above the ankle of an adult person with average height)
Question 2: Regardless of the height of the water, if others begin to cross the stream,
I will pass it, too.
Four central questions in this section were also quite similar to the first one, except that
the level of hypothetical stream water in each question was more than the previous one.
Given that the flow of water to a height of approximately 25 cm can cause a person to fall
down and sweep (Clements 2009), respondents who didn’t answer ‘‘agree’’ or ‘‘strongly
agree’’ to any of questions, considered as code 0 and others as code 1. Logistic regression
was used for statistical analysis in this study, and therefore, variables were coded in such
way (nominal dichotomous).
The second part of the questionnaire consisted of 10 questions to assess 10 independent
variables of study, i.e., the potential risk factors could be associated with risky behavior of
crossing the flood on foot. The independent variables included: age, gender, marital status,
level of education, attitude toward flood warning, experience expose to flood, experience
loss due to flood, perceived of flood threat, attitude toward level of swimming skills,
knowledge about consequences of flood. Since the optimal state to do logistic regression
can be achieved when the independent variables are nominal dichotomous, like dependent
variable, although different scales such as ordinal, nominal and interval were used to
measure independent variables, but these variables were considered as nominal dichoto-
mous and were coded as 0 and 1, then data entered into logistic regression model and, of
course, after applying Chi-square test.
As there is no specific formula to calculate the sample size for logistic regression
analysis, it is recommended that 15–20 samples per an independent variable should be
considered (Norman and Streiner 2000). According to the nine independent variables, and
with considering 10 % loss of the samples, the sample size was calculated as below:
15 � 10 ¼ 150; 150þ 15 ¼ 165
Target population in this study was residents over 18 years old of Quchan city that was
equivalent to over 18-years-old urban population auspices of the health network of Quchan
district. Noted population was 102,150 (2012).Of these, 165 persons were selected by
multi-stage sampling method. Seven Urban Health Centres (UHC) in Health Network of
Quchan district were considered as seven clusters. Considering the sample size and the
population coverage of each cluster, random sampling based on the last two digits of the
household health record number was performed. The inclusion criterion was the age of at
least 18 years old.
Data collection period was from April 24, 2013, to May 23, 2013. Researcher gathered
data by going directly to the homes of sample group; their addresses were obtained from
their health records of household. Researcher delivered the questionnaires and asked them
to return it, after completing.
As in this study, the dependent variable and the independent variables, both were
nominal dichotomous, at first, 2 9 2 cross tables were used to determine potential risk
factors associated with risky behavior of crossing the flood on foot. Then, significant
correlation between independent variables (potential risk factors) and the dependent var-
iable was examined by Pearson Chi-square test. Significance level of the test was con-
sidered B0.05 (P value B 0.05). Those independent variables that had a significant
Nat Hazards
123
correlation with dependent variable were entered into logistic regression analysis for
estimating predicting model of risky behavior of crossing the flood on foot.
3 Results
Totally, 159 subjects completed the questionnaire (response rate was around 96 %). The
relationships between possible individual risk factors and risky behavior of crossing the
flood on foot are described according to the results of 2 9 2 cross tables and results of
Pearson Chi-square test. Results of Pearson Chi-square test are reported in Table 1.
According to the results of 2 9 2 cross tables, 80 % of the age group 18–35 years were
positive to the question of whether or not they will pass through the flood on foot, while
only 60 % of age group 35? years had a positive response to the same question. As
Table 1 shows, the difference is statistically significant.
The percentage of positive responses to the question of passing through the flood on foot
was equal between men and women, respectively, 25 and 28 %. This minor difference was
not statistically significant (see Table 1).
The percentage of married respondents who say they would cross the flood on foot was
87 % married versus 70 % single, and the difference was almost significant (P = 0.053).
Sixty-seven percent of participants with a college degree say that they would pass
through the flood on foot, while the percentage of people with a lower education level that
stated would pass through the flood was 78. As shown in Table 1, the relationship between
level of education and risky behavior was not significant.
The percentage of people who took flood warning seriously, i.e., the ones state they
would leave immediately by alerts of receiving floods and cross the flood on foot, was 62
vs. to 87 % i.e., those who didn’t take flood warning seriously and say they would cross the
flood on foot. This difference was statistically significant (P = 0.001).
Eighty-one percent of respondents who did not have any experience of exposure to flood
say they would cross the flood on foot, which was higher than 61 % of those who say they
have had experience of exposure to flood and would pass through the flood on foot. This
difference was statistically significant (P = 0.022).
Percentage of respondents who had experience of loss due to flood and say they would
cross the flood on foot was close to percentage those respondents who had not experience
of loss due to flood and say they would cross the flood on foot (respectively, 79 and 72 %).
Certainly, the result of Chi-square test to examine the relationship between experience
losses due to floods and crossing the flood on foot was not significant.
The results showed percent of who had a proper perception of flood threat and say they
would cross the flood on foot was approximately equal to those who didn’t have a proper
perception of flood threat and say they would cross the flood on (75 vs. 72 %).
Seventy-eight percent of people with the right knowledge about immediate health
consequences of floods say they would cross the flood on foot, while the proportion of
people without right knowledge about immediate consequences of flooding on health that
state they would pass through the flood was 70 %. This difference was not statistically
significant.
In this study, 86 % of respondents who said their skill level in swimming were above
average, say they would cross the flood on foot, while only 66 % of those who stated their
swimming skills were in a elementary level or cannot swim, say they would pass through
the flood on foot and this difference was significant (P = 0.006).
Nat Hazards
123
As mentioned in the method, the independent variables that had a significant rela-
tionship with the dependent variable were entered into the logistic regression model as
probable predictors. These were four variables: age, attitude toward flood warning,
experience of exposure to flood and attitude toward level of swimming skills. The esti-
mated coefficients of the variables in the logistic regression model are presented in
Table 2. Hazmer and Lemeshow test indicated the values predicted by the model fitted to
the observed values (P = 0.0079).
According to the estimated odds ratios that are shown in the Table 2, the probability that
a person 18–35 years will cross the flood on foot (respond positively to question of sce-
nario of crossing the flood on foot) is 2.2 times than those over 35 years. The findings of
this table show that people who do not take seriously the warnings of flooding, two times
more likely than others to say they would cross the flood on foot. The odds ratio for the
variable of experience of exposure to flood is 1.948, i.e., the probability that people who
have experience of exposure to flood cross the flood on foot is almost half of those without
experience of exposure to flood. In the case of attitude toward level of swimming skills,
people who think they are in moderate to advanced level of swimming skills 2.3 times are
more likely to cross the flood on foot than those who think they are in a elementary level of
swimming skills or do not know swimming.
4 Discussion and conclusion
As expected, the percentage of the younger age group who say they would pass through the
flood on foot was more than the older age group. More experience and less level of
acceptable risk in elderly than youth could be the reason. Findings of a study conducted by
Ashley and Ashley (2008) were consistent with the findings of this study, as the death toll
from floods in the age group of 10–29 years were higher than in the older age groups.
Although it was expected the percentage of women who say they would cross the flood
on foot to be more than men, however, findings were contrary to expectations, as among
women are slightly higher. The results of a study conducted by Booth and Nolen (2012)
showed gender differences between boys and girls at the level of accepting risk at
uncertainty are not result of inherent differences between the sexes, but it can be a
reflection of social learning.
Table 1 Results of Pearson Chi-square test
Possible risk factor P value Chi square
Age 0.009 6.741
Gender 0.544 0.368
Marital status 0.053 3.743
Level of education 0.123 2.373
Attitude toward flood warning 0.001 10.417
Experience expose to flood 0.022 5.261
Experience loss due to flood 0.325 0.969
Perceived of flood threat 0.734 0.116
Attitude toward level of swimming skills 0.006 7.697
Knowledge about consequences of flood 0.231 1.435
Nat Hazards
123
It was expected married people to have lower level of accepting risk rather than single,
but percentage of married respondents who say they would cross the flood on foot was
larger than that of singles, and the difference was almost significant (P = 0.053). The
relationship between marital status and risky behavior of crossing the flood seemed to be
under the influence of other factors such as age or sex. Study of Molloy et al. (2009)
supports the finding. They indicated that after adjustment for age, sex and socioeconomic
status, the rate of death from cardiovascular disease (in fact, rate of high risk behaviors
leading to death from cardiovascular disease) was higher in single than in married, but
other studies show different results (Johnson et al. 2000; Ikeda et al. 2007). Generally, it
can be claimed that it depends on location and time.
The result of this study about relationship between level of education and risky behavior
is not consistent with the results of Geckova et al. (2002) Their finding shows a strong
correlation between level of education, as one of the three socioeconomic indicators and
some health risk behaviors such as smoking and alcohol consumption. However, the results
of that study indicated those with lower level of education are more likely to have risky
behaviors, which support our results.
As expected, the percentage of people who took flood warning seriously and cross the
flood on foot was lower than that of who didn’t take flood warning seriously and say they
would cross the flood on foot. As mentioned in the results, this difference was statistically
significant (P = 0.001) and was consistent with the findings of a research which investi-
gated the factors affecting driving behavior in floods. There was also a relationship
between flood warning attitude and behavior of driving into the flood (Drobot et al. 2007).
Lack of sufficient awareness and sensitivity toward the flood warning or warning of nat-
ural/manmade disasters can cause ignoring the warnings by some people.
The reason why people without any previous experience to deal with floods pass the
floods can be related to the fact that previous experience in exposure to a specific situation
or phenomenon result in more informed and more accurate decision making as this fact
proved in the study of Grothmann and Reussewing (2006).
In our study, there is no significant relationship between experience losses due to floods
and crossing the flood on foot that is consistent with the results of a research which
conducted by Drobot et al. (2007) they also could not prove any relationship between
experience loss due to flood and behavior of driving into the flood.
As the results showed percent of those who had a correct perception of flood threat and
say they would cross the flood on foot was approximately equal to percent of those who did
not have a correct perception of flood threat and say they would cross the flood (75 vs.
72 %). This was conflicting with our expectations. The relationship between these two
variables could be more depend on the time and place. Results of a study conducted by
Solis et al. (2009) support this reason.
The relationship between knowledge about consequences of flood and decision making
about crossing the flood on foot was not significant. It seems having knowledge about the
Table 2 Results of logistic regression
Variable b SE Wald Sig Odds ratio
Age 0.794 0.407 3.793 0.051 2.211
Attitude toward flood warning 1.008 0.457 4.864 0.027 2.740
Experience expose to flood 0.667 0.392 2.895 0.089 1.948
Attitude toward level of swimming skills 0.835 0.432 3.732 0.053 2.306
Nat Hazards
123
dangers of a particular phenomenon result in correct decisions, but when people want to
decide to do something (in our study, passing the flood or not), at first they analyze the
risks of that action. Slovic (2001) in his study concluded that risk is more than just knowing
the consequences and probability of consequences of an action. Considering such a defi-
nition of risk can help explain our result (in this case).
We use the level of swimming skills to determine whether respondents have ‘‘per-
ception of self-protection’’ or not. ‘‘The perception of self-protection’’ means an individual
believe he/she has the ability to save his/her health in a threatening situation. It is expected
if people face with a threatening situation have related ‘‘the perception of self-protection,’’
are more tend to deal with it than others. As it was expected proportion of respondents who
had ‘‘perception of self-protection’’ in floods (those who stated their skill level in swim-
ming is above average) and say they would cross the flood on foot was significantly more
than those who had not ‘‘perception of self-protection’’ in floods (those who stated their
swimming skills is in elementary level or cannot swim).
According to the results of logistic regression, four variables (age, attitude toward flood
warning, experience of exposure to flood and attitude toward level of swimming skills)
were determined as the four main predictors of risky behavior of crossing the flood on foot.
People with the age 18–35-years-old, those do not take flood warnings seriously, indi-
viduals who do not have experience of exposure to flood and those who believe they have
moderate to advanced level of swimming skills were identified as high risk groups in this
field, Accordingly, appropriate group training programs can be set for them to reduce
dangerous behavior of crossing the flood. Programs can be arranged to increase public
awareness on the importance and seriousness of flood warnings. It is suggested that
researches should be conducted to find solutions for the question of why some people do
not take the flood warnings seriously or to answer the question of why some people
crossing the flood on foot, despite the danger which threatens the people.
In addition, notification messages can be delivered to the public that swimming ability,
at any level, cannot help a person survive when he/she is being caught in the flood.
Based on our knowledge, we did not have any confounder for this research. The only
limitation for this research was that we had difficulty in separating those participants who
experienced the flood with those who did not have any experience.
Acknowledgments This research has been supported by School of Health Management and InformationSciences, Iran University of Medical Sciences.
References
(2012) Razavi Khorasan metrological office report of natural disasters. Retrieved 21 May 2013 from www.razavimet.ir
(2012) Vital Horoscope report of 2012 divided by districts. Retrieved 23 April 2012 from http://www.mums.ac.ir
(2014) Summarized table of natural disasters in Iran Islamic Rep from 1900 to 2014. The OFDA/CREDInternational Disaster Database. http://www.emdat.be/result-country-profile. Accessed 11 Jan 2014
Ardalan A, Holakouie Naieni K, Kabir MJ, Zanganeh AM, Keshtkar AA, Honarvar MR, Khodaie H, OsooliM (2009) Evaluation of Golestan province’s early warning system for flash floods, Iran, 2006–7. Int JBiometeorol 53:247–254
Ashley ST, Ashley WS (2008) Flood fatalities in the United States. J Appl Meteorol Climatol 47(3):805–818Booth AL, Nolen P (2012) Gender differences in risk behaviour: does nurture matter? Econ J 122(558):F56–
F78
Nat Hazards
123
Ciottone GR, Darling RG, Anderson PD, Auf Der Heide E, Jacoby I, Noji E, Suner S (2006) Disastermedicine. Mosby, Philadelphia
Clements B (2009) Disasters and public health: planning and response. Elsevier Inc, OxfordCoates L (1999) Flood fatalities in Australia, 1788–1996. Aust Geogr 30(3):391–408Coulston JE, Deeny P (2010) Prior exposure to major flooding increases individual preparedness in high-risk
populations. Prehosp Disaster Med 25(4):289–295Drobot SD, Benight C, Gruntfest EC (2007) Risk factors for driving into flooded roads. Environ Hazard
7(3):227–234Fillmore EP, Ramirez M, Roth L, Robertson M, Atchison CG, Peek-Asa C (2011) After the waters receded:
a qualitative study of university official’s disaster experiences during the Great Iowa Flood of 2008.J Community Health 36(2):307–315
FitzGerald G, Du W, Jamal A, Clark M, Hou XY (2010) Flood fatalities in contemporary Australia(1997–2008). Emerg Med Australas 22(2):180–186
Geckova A, Van Dijk JP, Groothoff JW, Post D (2002) Socio-economic differences in health risk behaviourand attitudes towards health risk behaviour among Slovak adolescents. Soz Praventivmed47(4):233–239
Grothmann T, Reussewing F (2006) Reussewing, people at risk of flooding: why some residents takeprecautionary action while others do not. Nat Hazards 38(2):101–120
Ikeda A, Iso H, Toyoshima H, Fujino Y, Mizoue T, Yoshimura T, Inaba Y, Tamakoshi A, JACC StudyGroup (2007) Marital status and mortality among Japanese men and women: the Japan CollaborativeCohort Study. BMC Public Health. doi:10.1186/1471-2458-7-73
Janerich DT, Stark AD, Greenwald P, Burnett WS, Jacobson HI, McCusker J (1981) Increased leukemia,lymphoma, and spontaneous abortion in Western New York following a flood disaster. Public HealthRep 96(4):350–356
Johnson NJ, Backlund E, Sorlie PD, Loveless CA (2000) Marital status and mortality: the national longi-tudinal mortality study. Ann Epidemiol 10:224–238
Jonkman SN, Kelman I (2005) An analysis of the causes and circumstances of flood disaster death. Disasters29(1):75–97
Kellar DMM, Schmidlin TW (2012) Vehicle-related flood deaths in the United States. J Flood Risk Manag5(2):153–163
Kreibich H, Thieken AH, Petrow T, Muller M, Merz B (2005) Flood loss reduction of private householdsdue to building precautionary measures—lessons learned from the Elbe flood in August 2002. NatHazards Earth Syst Sci 5:117–126
Molinari D, Handmer J (2011) A behavioural model for quantifying flood warning effectiveness. J FloodRisk Manag 4:23–32
Molloy GJ, Stamatakis E, Randall G, Hamer M (2009) Marital status, gender and cardiovascular mortality:behavioural, psychological distress and metabolic explanations. Soc Sci Med 69(2):223–228
Norman GR, Streiner DL (2000) Biostatistics: the bare essentials. B.C Decker Inc., London, p 119Schad I, Schmitter P, Saint-Macary C, Neef A, Lamers M, Nguyen L, Hilger T, Hoffmann V (2012) Why do
people not learn from flood disasters? Evidence from Vietnam’s northwestern mountains. Nat Hazards62:221–241
Seyedin H, Ryan J, Sedghi S (2011) Lessons learnt from the past and preparedness for the future: how adeveloping country copes with major incidents. Emerg Med J 28:887–891
Slovic P (2001) The risk game. J Hazard Mater 86(1–3):17–24Solı́s D, Thomas M, Letson D (2009) Determinants of household hurricane evacuation choice in florida. In:
The southern agricultural economics association annual meeting, Atlanta, GeorgiaSorensen JH (2000) Hazard warning systems: review of 20 years of progress. Nat Hazards Rev 1:119–125Tapsell SM, Tunstall SM (2008) I wish I’d never heard of Banbury: the relationship between place and the
health impacts from flooding. Health Place 14(2):133–154Tapsell SM, Penning-Rowsell EC, Tunstall SM, Wilson TL (2002) Vulnerability to flooding: health and
social dimensions. Philos Trans A Math Phys Eng Sci 15:1511–1525
Nat Hazards
123