Study designs
An analytical cross-sectional study was conducted with a sample of 316 students belonging to the only three secondary education institutions in Muisne, Ecuador. The data was collected during January and February 2017, nine months after the earthquake.
Settings
This study was completed in Muisne, a coastal town in the province of Esmeraldas in Ecuador. Ecuador is a country located in Southamerica sharing borders with Peru and Colombia with 16,623.000 inhabitants. The Esmeraldas province has 7 cantons including Muisne, being this canton, a beach town located an hour and a half from the capital of the province.
The coastal town of Muisne runs along 8 kilometers of coast and has an estimated population of 28, 047 inhabitants, being 50.03% women and 49,97% men, most of them afro-Ecuadorians.
This part of the country was heavily affected by the 2016 earthquake and subsequent aftershocks.
Participants
We included a sample of 316 students, 145 men (48.1%) and 156 women (51.9%) young adolescents (13 to 19 years of age) with an average age of 16.07 years (SD = 1.91).
Eligibility criteria
The study formally enrolled students within the education system. All the subjects that voluntarily agreed with the informed consent and their parents authorized their participation were included.
Variables
Socio-demographic variables such as age, sex, and level of education were analyzed, as well as the variables of interest regarding earthquake-related-stressors, Post-traumatic Stress Disorder, anxiety, depression and suicidal risk.
Post-traumatic Stress Disorder Symptoms
The Child PTSD Symptom Scale (CPSS) was created based on previously published studies (24–27), to assess the presence of PTSD symptoms in children and adolescents between 8 and 18 years in terms of the last two weeks.
In the present study, we used the Spanish version translated by the Chilean researchers Bustos, Rincón and Aedo, instrument validated in 2009 and adapted to the Ecuadorian context and the situation(28). The answers may vary from: 0 (never or only once), 1 (occasionally), 2 (half the time) and 3 (almost always). For the present study, the 17 items of its original version were used. The possible scores vary from 0 to 51 and the cut-off point was established at 24 points (24). The scale showed high internal consistency and good test-retest reliability in its original version (24) as in the Chilean adaptation (28).
Anxiety
The Spence Children's Anxiety Scale, (SCAS) was used to evaluate the presence of different anxiety disorders among children and adolescents(29). The SCAS provides a description of the severity of specific anxiety symptoms for both children and young adolescents (30) and is characterized by measuring the specifications of anxiety disorders in children and adolescents. For this work, a validated version in Spanish was taken (31,32), with proven reliability and validity (33,34).
In order to shorten the questionnaire, we used only three subscales (21 items) of original 38 items, which evaluated the types of anxiety directly linked to natural disasters: panic attacks, separation anxiety and generalized anxiety disorders. The internal consistency of the mentioned subscales varies from 0.81 (panic attacks) to 0.75 (separation anxiety) and 0.67 (generalized anxiety); Cronbach's alpha for the total scale is 0.89. Since there is no cut-off point for these three subscales in this study, the degree of affectation was determined considering one standard deviation from the media that in the present case was 35 points.
Prevalence of depression
The Center for Epidemiologic Studies Depression Scale (CES-D) scale was designed to determine prevalence of depression in adolescents and adults in a general population (35). The inventory evaluates the presence of twenty depressive symptoms during the last week, with a response format of: No day = 0, one to three days = 1, four to six days = 2 and every day = 3.
The scale has been validated in Mexican and Chilean students (36), Colombians (37) (32)and Peruvians (38), with a good result in terms of reliability and validity. In the present study, the version used among Ecuadorian adolescents presents a Cronbach's alpha of 0.81. We used the cut-off point of 24, methodology validated by Fuentealba et al, 2004 in a similar study among youths from Chile, in Southamerica(36).
Suicidal risk
The Okasha Suicidality Scale was created by Okasha et al in 1981 and was used to determine the risk of committing suicide within this community(39). This scale measures suicide risk and consists of 4 items: 1. Have you ever thought that life is not worth it? 2. Have you ever wanted to be dead? 3. Have you ever thought about ending your life? 4. Have you attempted suicide?
The first three items have a 0 to 3 format as a response option (never, almost never, sometimes, many times). In the present study a fifth item was added: Have you made any plans to take your life?, since the plan plays an important role for the assessment of suicide risk (40).
The original scale was translated into Spanish and validated by Salvo et al. 2009 in a teenage population in Chile with an internal consistency of 0.89(41). The cut-off point for the subscale of suicidal ideations, to identify people with high suicidal tendencies, was set at 5 points (41).
Earthquake-related stressors
An adapted seven-questions survey based on the study conducted by Díaz, Quintana and Vogel was used to explore the earthquake-related stressors (42). The questionnaire contained seven items that were intended to identify the earthquake-related stressors that might be linked to mental health problems or suicide risk after an earthquake. The questions were intended to explore the timeline related to the earthquake, the degree of damage in houses and the consecutive economic losses, the current housing and the housing just after the earthquake, the physical damages and life losses affecting their selves and their relatives and the psychological support (if any) that was available just after the disaster.
The social determinants (earth-quake related stressors) were dichotomized. We decided to do so because the items included several categories, starting from very mild affections to seriously affecting stressors. Thus, we decided to purposefully dichotomized them in no affection and affections. After doing this, we were able to complete binomial analysis using a linear regression model.
Data quality and validation
Missing data was replaced by the average of each participant, value produced after calculating the mean after completing the scale. For reasons of accuracy, no missing was replaced if they represented more than 20% of the scale.
Bias
In this study we try to include all adolescents who voluntarily decided to participate, so we could avoid selection bias. We have considered the subsequent analysis of the data to try to avoid design bias to the maximum and we also recognized that some information from the past might cause some degree of memory bias, situation that we expand within the limitation section.
Study Size
To get a representative sample of the adolescent population of Muisne, we collected the data from the only three existing schools (two public and one private) in Muisne. The Ministry of Education’s official report included 1002 students within the ninth to twelfth grade for the 2016-2017 period.
We calculated the formula for the ideal sample size as follows: n = (Z2 x P (1 - P)) / e2). Where Z corresponds to the value of the standard distribution based on the confidence level of 0.05% while the P indicates the expected prevalence (in this case the highest expected prevalence, based on the study by Diaz, Quintana & Vogel (2012) is 30%) and e corresponds to the expected accuracy. Adding the student population of the three schools (N = 1002) we obtained an optimal sample of 321 participants. The voluntary participation percentage in each selected classroom was 50%, so 319 questionnaires were collected. After reviewing the questionnaires, only those with complete information were (301 participants).
Statistical analysis
For the two groups comparisons we used a t-test in order to compare means obtained in the measured scales (suicidal tendency and other mental health events between the samples).
In terms of the earthquake-related stressors we dichotomized them according to the presence or not of any effect (0 if the damage has been slight or nonexistent and 1 if it has been serious). To contrast the dichotomized and the demographic variables (gender and age) a Pearson Chi-square test was used. A correlation and a linear regression analyses were run using the sum of all the earthquake-related stressors in order to estimate the impact of various stressors on mental health and suicidality tendencies.
Lineal regression was carried out to test the predictive power of depression, anxiety, PTSD and earthquake related stressors for suicidal tendency (table 4). These factors explain altogether 25% of variance in suicidality, with highest beta coefficients for depression (β=.30, p=.003), generalized anxiety disorder (β=.20, p=.013), gender (β=.20, p=.003), posttraumatic stress disorder (β=.19, p=.025) and at least earthquake related stressors (β=.16, p=.013).
The three subscales were analyzed separately: panic attacks, separation anxiety, and generalized anxiety disorders.All statistical analysis accepted significance with an alpha level of 0.05%. The statistical analysis was completed using the IBM SPSS statistics version 24.0. while all the references were managed using the Zotero Open Source Software Version 5.0.84, 2020.