2.1. Sample
The study included 780 (mean age 26.56 ± 13.21; age range 18–84 years) California residents sampled in Spring 2019, six-months after the 2018 Camp Fire. Participants were recruited at three sites in California, one in San Diego, and two in Chico. While San Diego is at a distance of approximately 600 miles from the center of the Camp Fire, Chico is one of the cities whose residents were most affected, within 10–15 miles of the center of the Camp Fire. The samples recruited from Chico were either students in the department of psychology at the California State University (CSU) (sample n = 417) or individuals enrolled in the CSU Basic Needs program that provided disaster relief and community-based support directly to Camp Fire victims (sample n = 110) (39). The sample from San Diego was recruited at the University of California San Diego (UCSD, sample n = 253). The response rate of invited study participants from the CSU Basic Needs program was 37%; response rates could not be determined for CSU Psychology and UCSD samples as there was no determinable upper limit to the number of individuals who may have seen the recruitment advertisement. Based on these recruitment sites in relation to the Camp Fire, we refer to these groups as “primary proximity and help seeking” (those in the Basic Needs program), “primary proximity” (CSU students not in the Basic Needs program), “secondary proximity” (those at San Diego).
The study was approved by the institutional review boards (IRB) of the University of California San Diego and California State University at Chico. All study participants provided written informed consent. A sample comparison regarding demographic variables from the three recruitment sites is shown in Table 1.
Table 1
Sample demographics across all study sites
|
|
Primary proximity
help seeking
M ± SD
n (%)
|
Primary proximity
M ± SD
n (%)
|
Secondary proximity
M ± SD
n (%)
|
χ2
|
p
|
Age
|
|
27.35 ± 9.74
|
21.60 ± 3.42
|
34.40 ± 19.41
|
107.20
|
< .001
|
Sex
|
Male
Female
|
28 (25.5)
82 (74.5)
|
85 (20.4)
332 (79.6)
|
102 (40.3)
151 (59.7)
|
31.62
|
< .001
|
Ethnicity
|
Caucasian
African American
Asian
mixed/other
n/a
|
77 (70.0)
3 (2.7)
7 (6.4)
13 (11.8)
10 (9.1)
|
238 (57.1)
16 (3.8)
19 (4.6)
64 (15.3)
80 (19.2)
|
130 (51.4)
3 (1.2)
72 (28.5)
30 (11.9)
18 (7.1)
|
104.50
|
< .001
|
SES
|
|
1.75 ± .68
|
2.11 ± .74
|
2.27 ± .72
|
40.03
|
< .001
|
Note. n/a = unknown or not reported, SES = socio-economic status (range 0–9). χ2 = Chi-Square statistics derived from non-parametric group comparisons. |
2.2. Measures
All study participants reported on the following measures at six-months after the 2018 Camp Fire. This time of assessment is considered suitable in signifying chronic mental health outcomes (40). The Research Electronic Data Capture (REDCap) tool was used for survey administration.
Demographics. Assessed demographic variables included age, sex, race (Caucasian, African American, Pacific Islander, Asian, American Indian or mixed), years of education and socio-economic status (SES). Due to low prevalence of individuals with American Indian or Pacific Islander origin (< 1%), we extended the racial category “mixed” to “mixed/other”. SES composite scores were assessed using the family affluence scale; this scale measures family wealth based on family ownership of objects of value (e.g., car/computer) and produces a composite score ranging from 0 (low affluence) to 9 (high affluence) (41). Demographic predictors that have previously been associated with effects of environmental disasters on mental health include sex and SES (17).
Life events. We evaluated life events related to fire and to childhood adversity. For the former, we used the Life Events Checklist (LEC-5; (42)) from the Diagnostic and Statistical Manual of Mental Disorders (DSM-5), which assesses potentially traumatic events that might have happened at any time in life. It inquires the exposure to events on a 5-point nominal scale (happened to me = 1, witnessed it = 2, learned about it = 3, part of my job = 4, not sure = 5, does not apply = 0). The events cover a vast variety, yet we focused on an item assessing exposure to fire. Positive responses can be given to multiple levels of exposure, if applicable. Note that LEC-5 “witnessed it” responses on fire exposure pertain to real life events and do not include exposure to pictures or videos of the fire in the media. Note that we did not explicitly inquire about exposure to the 2018 Camp Fire as the CSU IRB objected to inclusion of such a direct question given its potential to lead to further traumatic stress.
Additionally, experiences of child maltreatment during the first 18 years of life were assessed using the 28-items brief screening version of the Childhood Trauma Questionnaire (CTQ; (43)). The CTQ is a self-administered retrospective inventory consisting of five categories of child neglect and abuse. Each category entails five items that are rated on a 5-point Likert scale ranging from “never true” (= 1) to “very often true” (= 5).
Mental health outcomes. All mental health scales for PTSD, depression and anxiety were assessed in self-report. PTSD symptom severity was measured using the PTSD-Checklist (PCL-5 (44)). The PCL-5 consists of 20 questions that cover experience of PTSD symptoms such as memories, dreams, avoidance of certain external or internal stimuli etc. Responses are given on a 5-point Likert scale asking how much one was bothered by these experiences from “not at all (= 0)” to “extremely (= 4)”. To assess major depressive disorder (MDD), we used the Patient Health Questionnaire (PHQ-9; (45)), a diagnostic instrument that scores each of the 9 DSM-5 criteria for MDD on a 4-point Likert scale assessing frequency of symptoms from “not at all” (= 0) to “nearly every day” (= 3). To assess Generalized Anxiety Disorder (GAD), we used the 7-item brief scale GAD-7 (46). Frequency of anxiety symptoms is scored on a 4-point Likert scale from “not at all” (= 0) to “nearly every day” (= 3).
Resilience factors. We assessed subjective resilience and resilient lifestyle factors, specifically, (I) sleep quality, (II) exercise, (III) mindfulness and (IV) emotional support.
The Brief Resilience Scale (BRS; (47)) measures resilience as the capacity to bounce back after tough times, with 6 items on a 5-point Likert scale, half of them inverted from “strongly disagree” (= 1) to “strongly agree” (= 5).
(I) Sleep quality was assessed with regards to sleep disturbances, using the short form of the Patient-Reported Outcomes Measurement Information System (PROMIS) - Sleep Disturbance scale (48). It consists of 8 items that assess sleep disturbances (e.g. “my sleep was restless”) in the past week on a 5-point Likert scale ranging from “not at all” (= 1) to “very much” (= 5).
(II) To investigate physical exercise, we use three questions of the Godin Leisure-Time Exercise Questionnaire (49). Participants are asked how many times on average during a typical 7-day period, they do strenuous, moderate or mild exercise for more than 15 minutes.
(III) The disposition of being mindful, which can be conceptualized as open and receptive awareness, was measured with the Mindful Attention Awareness Scale (MAAS; original scale by (50), implemented in the version described by (51)). The MAAS entails 14 items such as “I could be experiencing some emotion and not be conscious of it until sometime later.” that are inversely scored on a 6-point Likert scale from “almost always” (= 1) to “almost never” (= 6).
(IV) Due to its well documented impact on mental and physical health, we additionally measure emotional support from social relationships using the respective subscale from the NIH Toolbox on Social Relationships (SR) (52). The sum of 8 items on a 5-point Likert scale from “never” (= 0) to “always” (= 4) represents the presence and frequency of social support in the participants’ lives.
2.3. Data Analysis
Exposure to wildfires across participant subgroups.
We compared the participant subgroups recruited from the three sites in their self-reported degree of exposure to the wildfires. For this, we used the “fire or explosion” life event of the LEC-5 and computed chi-square tests for all five qualitatively different forms of exposure between the three participant pools. As the study sample was recruited from different locations in California, this analysis step is included to confirm that study participants in closer proximity to the outbreak of the fires were indeed more directly exposed.
Mental health outcomes of wildfires.
To study mental health outcomes of the wildfires, we computed separate multiple regressions for each mental health outcome. Group membership (directly exposed, indirectly exposed, not exposed) together with demographic variables age, sex, ethnicity and SES were modeled as predictors. Due to non-parametric score distributions in the mental health outcomes, we performed bootstrapping using 1000 bootstrap samples. We report two statistical measures of model fit. The F-test for overall model significance indicates whether the regression model fits the data, and R2 indicates the percentage of explained variance of the outcome.
Using hierarchical multiple regression analyses, factors of vulnerability and resilience (i.e., childhood trauma, resilience, sleep quality, exercise, mindfulness and emotional support) were modeled as additional predictors. This approach allowed to test whether vulnerability and resilience factors add value to the prediction of mental health outcomes, as indicated by a significant increase in R2 by means of model extension.