ORIGINAL RESEARCH

published: 12 January 2022

doi: 10.3389/fpsyt.2021.752870

Vulnerability and Protective Factors

for PTSD and Depression Symptoms
Among Healthcare Workers During
COVID-19: A Machine Learning
Approach
Liana C. L. Portugal 1,2 , Camila Monteiro Fabricio Gama 2 , Raquel Menezes Gonçalves 2 ,
Mauro Vitor Mendlowicz 3 , Fátima Smith Erthal 4 , Izabela Mocaiber 5 , Konstantinos Tsirlis 6 ,
Eliane Volchan 4 , Isabel Antunes David 2 , Mirtes Garcia Pereira 2† and Leticia de Oliveira 2*†
1
Neurophysiology Laboratory, Department of Physiological Sciences, Roberto Alcantara Gomes Biology Institute, Biomedical
Center, State University of Rio de Janeiro, Rio de Janeiro, Brazil, 2 Laboratory of Neurophysiology of Behavior, Department of
Physiology and Pharmacology, Biomedical Institute, Federal Fluminense University, Rio de Janeiro, Brazil, 3 Department of
Psychiatry and Mental Health, Fluminense Federal University, Rio de Janeiro, Brazil, 4 Laboratory of Neurobiology, Institute of
Edited by: Biophysics Carlos Chagas Filho, Rio de Janeiro, Brazil, 5 Laboratory of Cognitive Psychophysiology, Department of Natural
Tingzhong Yang, Sciences, Institute of Humanities and Health, Federal Fluminense University, Rio de Janeiro, Brazil, 6 Centre for Medical Image
Zhejiang University, China Computing, University College London, London, United Kingdom
Reviewed by:
Bing Zhang,
Nanjing Drum Tower Hospital, China
Background: Healthcare workers are at high risk for developing mental health problems
Fengqin Wang, during the COVID-19 pandemic. There is an urgent need to identify vulnerability and
Hubei Normal University, China
protective factors related to the severity of psychiatric symptoms among healthcare
*Correspondence:
workers to implement targeted prevention and intervention programs to reduce the
Leticia de Oliveira
oliveira_leticia@id.uff.br mental health burden worldwide during COVID-19.
† These authors have contributed Objective: The present study aimed to apply a machine learning approach to predict
equally to this work and share last depression and PTSD symptoms based on psychometric questions that assessed: (1)
authorship
the level of stress due to being isolated from one’s family; (2) professional recognition
Specialty section: before and during the pandemic; and (3) altruistic acceptance of risk during the COVID-19
This article was submitted to pandemic among healthcare workers.
Computational Psychiatry,
a section of the journal Methods: A total of 437 healthcare workers who experienced some level of isolation
Frontiers in Psychiatry at the time of the pandemic participated in the study. Data were collected using
Received: 03 August 2021 a web survey conducted between June 12, 2020, and September 19, 2020. We
Accepted: 08 November 2021
Published: 12 January 2022
trained two regression models to predict PTSD and depression symptoms. Pattern
Citation:
regression analyses consisted of a linear epsilon-insensitive support vector machine
Portugal LCL, Gama CMF, (ε-SVM). Predicted and actual clinical scores were compared using Pearson’s correlation
Gonçalves RM, Mendlowicz MV,
coefficient (r), the coefficient of determination (r2 ), and the normalized mean squared error
Erthal FS, Mocaiber I, Tsirlis K,
Volchan E, David IA, Pereira MG and (NMSE) to evaluate the model performance. A permutation test was applied to estimate
Oliveira Ld (2022) Vulnerability and significance levels.
Protective Factors for PTSD and
Depression Symptoms Among Results: Results were significant using two different cross-validation strategies to
Healthcare Workers During COVID-19: significantly decode both PTSD and depression symptoms. For all of the models, the
A Machine Learning Approach.
Front. Psychiatry 12:752870. stress due to social isolation and professional recognition were the variables with the
doi: 10.3389/fpsyt.2021.752870 greatest contributions to the predictive function. Interestingly, professional recognition

Frontiers in Psychiatry | www.frontiersin.org 1 January 2022 | Volume 12 | Article 752870

Portugal et al. PTSD and Depression During Covid-19

had a negative predictive value, indicating an inverse relationship with PTSD and
depression symptoms.
Conclusions: Our findings emphasize the protective role of professional recognition
and the vulnerability role of the level of stress due to social isolation in the severity of
posttraumatic stress and depression symptoms. The insights gleaned from the current
study will advance efforts in terms of intervention programs and public health messaging.

Keywords: COVID-19, PTSD, depression, healthcare worker (HCW), machine learning

INTRODUCTION which can be explained in part by the additional social, structural

and political problems in Brazil (16).
Coronavirus disease 2019 (COVID-19) is an infectious disease Developing strategies to protect mental health, especially in
caused by the novel coronavirus (SARS-Cov2). In March 2020, this population, is an important task for governments and health
the World Health Organization (WHO) characterized COVID- systems around the world (17), especially in countries with great
19 as a pandemic due to the rapid increase in the number inequalities in income/wealth, such as Brazil (18). An important
of cases, putting the planet in a state of maximum alert (1). step is to identify vulnerability and protective factors to prevent
Driven by an infectious new variant, a lack of containment mental disorders from progressing (19), which becomes even
measures and a patchy vaccine rollout, Brazil has become the more relevant and challenging when applied in the context of the
epicenter of the COVID-19 pandemic. According to the most COVID-19 pandemic.
recent WHO estimates, Brazil has the highest numbers of new Insights about vulnerability and protective factors that impact
deaths in the Americas (2). During the period of our research, the mental health of healthcare workers during the COVID-19
Brazil surpassed 4.5 million COVID-19 cases, and more than pandemic have already been provided by studies investigating
136,000 Brazilians have died from COVID-19 since the start of many objective aspects, such as years of work, professional level,
the pandemic (3). At the time of the research, an effective vaccine gender and age (20–25). Here, we focused on aspects about the
or medicine was not available to address COVID-19, and the self-perception of daily professional life in dealing with COVID-
most efficient strategies for controlling the COVID-19 pandemic 19 that are still relatively unknown, such as the perception of
were preventive measures and social distancing. According to stress from being isolated, professional recognition, and altruistic
an article in Lancet, Brazil was considered to have had one of acceptance of risk.
the worst responses to the pandemic internationally and to have In line with this notion, for example, most studies showing
committed numerous governmental mistakes (4). negative associations between social isolation and mental health
In this context, the COVID-19 pandemic not only raises outcomes during the COVID-19 pandemic have evaluated
physical health concerns for the entire population but also objective aspects of isolation (e.g., duration of isolation, local
has consequences on the mental health of individuals in structure in which isolation occurs and comparisons of the
both the short and long terms (5–8), particularly among mental health outcomes of individuals who were isolated from
healthcare workers, a group with higher risks of infection and of those who were not) (26–29). Furthermore, studies exploring
transmitting the disease to their families and coworkers (9, 10). the psychological impact of social isolation in healthcare
In fact, studies from previous epidemics, such as SARS, Ebola professionals during COVID-19 remain scarce, and the role of
and MERS, have shown that healthcare workers are vulnerable to self-perceived level of stress from being isolated from one’s family
mental health problems (10–12) and that some consequences can in predicting psychiatric symptoms remains undetermined. It is
be persistent (11). In the current COVID-19 pandemic, a recent important to emphasize here that, although subjective feelings
systematic review and meta-analysis showed a high prevalence of social isolation and the objective state of social isolation
of depression (31.1%) and posttraumatic stress disorder (PTSD; frequently co-occur, studies have suggested that they are not
31.4%) among caregivers in practice worldwide (13). PTSD is a equal; both can exert a detrimental effect on health through
mental health problem that affects people who are exposed to shared and different pathways (30).
potentially traumatic events. In particular, healthcare workers are In general, much less attention has been given to factors
vulnerable to PTSD because they are directly exposed to COVID- that could be associated with protection against poor mental
19 trauma, including the death of patients due to COVID- health outcomes during an epidemic, including self-perceived
19, danger of contamination and the possibility of transmitting professional recognition and altruistic acceptance of risk.
SARS-Cov2 to another person (14, 15). However, until now, According to findings from previous epidemics, professional
no studies have investigated symptoms of PTSD for traumas recognition can be considered a motivating factor for medical
specifically related to COVID-19 in healthcare professionals. teams to continue working in future epidemics (31). In the
Regarding Brazil, a web survey conducted at the beginning of the current COVID-19 pandemic, professional recognition has
COVID-19 pandemic showed that living in Brazil was associated emerged as a significant protective factor against burnout
with increased odds of depression among essential workers, syndrome (32); however, it is necessary to expand knowledge

Frontiers in Psychiatry | www.frontiersin.org 2 January 2022 | Volume 12 | Article 752870

Portugal et al. PTSD and Depression During Covid-19

to other psychiatric conditions, such as PTSD and depression. METHODS AND MATERIALS
Altruistic intent to help, a quality frequently found among
healthcare workers, was related to a statistically significant Study Design and Recruitment Procedure
decrease in PTSD and depression symptoms during the SARS This study was part of a broader project, the PSIcovidA project,
outbreak among hospital employees in Beijing, China (11, 12). aimed at investigating the impact of traumatic events related to
While altruistic intent to help has been shown to be protective the COVID-19 pandemic on professionals working in hospital
against psychiatric symptoms in previous epidemics, no studies environments or in emergency care units acting directly or
have assessed the role of perceived altruistic acceptance of risk in indirectly in the fight against the COVID-19 pandemic in
the prediction of depressive and PTSD symptom severity levels Brazil. PSIcovidA has a cross-sectional data and follow-up survey
during the COVID-19 pandemic. design. This paper presents cross-sectional that were collected
Identifying vulnerability and protective factors for mental over 3 months between June 12, 2020, and September 19, 2020.
health is a major challenge in psychiatry. Currently, we Data were collected by a convenience snowball sampling
can apply artificial intelligence, such as machine learning technique from professionals working in different healthcare
approaches, to find individual predictions that can help contexts or in emergency care units in different states of
to detect mental health vulnerabilities (33–35). Machine Brazil. An online survey was developed and sent by WhatsApp
learning is a rapidly emerging field that has the potential Messenger (WhatsApp Inc, Mountain View, CA, USA) and e-
to identify multivariate patterns in psychometric data that mail. An Instagram account and a webpage for the PSIcovidA
enable the classification of an independent series of individuals project were created to advertise the project. Furthermore, the
(classification model) or the prediction of continuous variables professional associations of all major healthcare worker groups
(such as symptoms) at the individual subject level (pattern in Brazil were contacted to publish the main project proposal
regression model). In fact, pattern regression models could allow and the link to complete the survey online on their websites and
for the investigation of mental health outcomes that represent on Instagram. Moreover, interviews in Brazilian media about the
vulnerability to or protection against the severity of psychiatric study were conducted to invite people who worked in hospitals
symptoms. However, there are still few studies using pattern or emergency units to participate.
regression models to predict mental health symptoms based on Participants were asked to complete a set of validated
psychometric data during the COVID-19 pandemic (37–39). questionnaires that included sociodemographic questions, as well
Here, we aimed to apply pattern regression models based on as questions about professional recognition before and during
psychometric data to predict depression and PTSD symptoms the pandemic; mental disorder symptoms, including symptoms
among healthcare workers during the COVID-19 pandemic. of depression and PTSD; social isolation from one’s family; and
A fundamental insight from the field of statistical learning altruistic acceptance of risk. At the end of the questionnaires,
is that the ability of a model to predict the values of new participants were presented with a list of online psychological
observations will generally be overestimated based on the fit of support groups.
the model to a particular dataset [(39); for a review, see (40)]. This study was approved by the Ethics Research Committee
In the context of machine learning, the term “predict” means of Federal Fluminense University (UFF) and National Research
that, once the model has learned a relationship between a set Ethics Commission (CONEP) under process number CAAE
of patterns (e.g., multivariate patterns of psychometric data) 31044420.9.0000.5243, and all of the participants agreed to
and labels (e.g., a clinical score), given a new pattern (e.g., participate voluntarily in the survey.
psychometric data from a new subject), it can predict its label.
Despite being innovative, the advantages of this method include Participants
the following: (1) models are not constrained by traditional In total, 1,843 respondents accessed the web survey and
assumptions, such as a normal distribution of the data or an a completed it. The inclusion criterion was being a hospital
priori model; (2) the method can evaluate relationships among and/or emergency health care worker, which generated a
many variables at once; and (3) it is particularly helpful for sample of 1,399 participants. The exclusion criteria included not
finding patterns in complex datasets (41). having experienced a traumatic event related to the COVID-
In summary, the present study aimed to apply a machine 19 pandemic situation (n = 220) or having failed to fully
learning approach (pattern regression model) for the first time complete the questionnaire battery (n = 178). Furthermore, 564
to predict depression and PTSD symptoms regarding traumatic participants who had not experienced some level of isolation,
events specifically related to the pandemic based on self- i.e., physical distance from one or more family members, such
perceived (1) level of stress from being physically isolated as children, brothers, husbands or wives, for at least 1 week at the
from one’s family; (1) professional recognition before and after time of the pandemic were also excluded. After the application
the pandemic; and (1) altruistic acceptance of risk during the of these criteria, the final sample consisted of 437 respondents
COVID-19 pandemic among hospital and/or emergency care representing all 26 states in Brazil. The majority of our sample
unit employees. There is an urgent need to identify vulnerability consisted of women (n = 320, 73.2%), 20-72 years (M = 39.5;
and protective factors for mental health, especially for healthcare SD = 10.8, range: 20-72 years), and a large proportion of the
workers, to implement targeted prevention and intervention respondents lived in the state of Rio de Janeiro (62%). The sample
programs to reduce the psychiatric burden affecting healthcare mirrored the Brazilian population of healthcare workers in terms
systems worldwide during the COVID-19 pandemic. of gender. Estimates by the National Council of Municipal

Frontiers in Psychiatry | www.frontiersin.org 3 January 2022 | Volume 12 | Article 752870

Portugal et al. PTSD and Depression During Covid-19

TABLE 1 | Sociodemographic and occupational characteristics of the participants. protective factors for mental health disorders (PTSD and
depression symptoms).
N (%); Mean (SD)

Sociodemographic characteristic
Professional Recognition
Gender
Respondents were asked to rate their perceived professional
Female 320 (73.2%)
recognition before and during the pandemic using a 10-point
Male 117 (27.8%) Likert scale. In particular, they were asked to answer the following
Age 39.5 (10.8) question: “In your opinion, from 1 to 10, how much did the
Professional level general population appreciate healthcare professionals?” (1 = not
Technician 87 (19.9%) at all, 10 = too much).
Superior 350 (80.1%)
Profession
Altruistic Acceptance of Risk
Medical doctor 173 (39.6%)
The item “Because I wanted to help the COVID-19 patients,
Nurse 72 (16.5%)
I was willing to accept the risks involved” was used as a
Nurse technician 60 (13.7%)
measure of altruistic acceptance of risk. Respondents were asked
Physiotherapist 43 (9.8%) to rate this item from 1 (not at all) to 10 (extremely true).
Clinical psychologist 27 (6.2%) This question was adapted from the 10th item of the Perceived
Pharmacist 19 (4.4%) Threat Questionnaire developed by Chong et al. during the SARS
Other 43 (9.8%) pandemic (43).
Region
Southeast 321 (73.5%)
Stress Due to Social Isolation
South 34 (7.8%)
The respondents were asked to rate their level of stress due to
North 18 (4.1%)
being isolated from one or more members of their families for at
Northeast 57 (13.0%) least 1 week at the time of pandemic using a 10-point rating scale
Midwest 7 (1.6%) (1 = low, 10 = high).
Institution
Public 228 (52.2%) Variables to Be Predicted—Psychometric
Private 86 (19.7%) Scales for PTSD Symptoms and
Both 123 (28.1%) Depression Symptoms
Presence of mental disorder Posttraumatic Stress Disorder Checklist 5
No 309 (70.7)
Posttraumatic stress symptoms were assessed using the PCL-
Yes 128 (29.3)
5, which was developed by the National Center for PTSD in
Worst trauma Covid
accordance with the DSM-5 criteria (44, 45). This scale was
Learning about the death of a close relative or 94 (21.5%)
translated and adapted to Portuguese by Lima et al. (46). The
coworker
PCL-5 is a 20-item self-report questionnaire that measures four
Possibly transmitting the COVID-19 virus to 90 (20.6%)
another person clusters of symptoms of PTSD: intrusion, avoidance, negative
Experiencing the imminent risk of death of a 72 (16.5%) alterations in cognition and mood, and alterations in arousal and
close relative or coworker reactivity. Each item on the PCL-5 questionnaire is assessed via
Personally witnessing the death of a patient 67 (15.3%) a five-point Likert scale (from 0 = not at all to 4 = extremely).
Being infected with COVID-19 48 (11.0%) Symptom severity can be calculated by totaling the items for each
Being exposed to infected patients at high risk 47 (10.8%) of the four clusters or totaling all 20 items; in this case, the severity
for death score ranged from zero to 80 points.
Personally witnessing the death of a close 19 (4.3%) The participants were instructed to complete the PCL-
relative or coworker
5 in relation to their worst traumatic experience related
to the COVID-19 pandemic. To assess the worst trauma,
we developed a questionnaire composed of seven items
Health Secretariats (CONASEMS), based on Brazilian Institute
that investigated traumatic situations experienced during the
of Geography and Statistics (IBGE) data, indicate that women
COVID-19 pandemic and the level of stress associated with
represent 65% of the more than six million professionals working
them. These situations included (1) personally witnessing the
in the public and private health sectors at all levels of care
death of a patient due to COVID-19; (2) personally witnessing
complexity (42). Additional sociodemographic information is
the death of a family member or coworker due to COVID-
presented below (Table 1).
19; (3) learning, through others, about the death of a family
member or a coworker due to COVID-19; (4) experiencing
Predictive Variables—Psychometric the imminent risk of death of a family member or coworker
Questions due to COVID-19; (5) being exposed to critically ill patients
The pattern regression models included three different infected with COVID-19 whose lives were in danger; (6)
psychometric questions assessing vulnerability and being infected with COVID-19; and (7) believing or having

Frontiers in Psychiatry | www.frontiersin.org 4 January 2022 | Volume 12 | Article 752870

Portugal et al. PTSD and Depression During Covid-19

TABLE 2 | The means and standard deviations for the psychometric questions training phase, the model was trained by providing examples
and the scales in the considered sample. of psychometric questions (i.e., professional recognition before
Factor Variable Mean (SD)
and during the pandemic, altruistic acceptance of risk and
stress due to social isolation) and a label (variables to be
Protective factors predicted: posttraumatic or depression symptoms). Once the
Professional recognition (before the pandemic) 4.4 (1.9) model “learned” the association between the question scores and
Professional recognition (during the pandemic) 7.2 (2.0) the label from the training data (i.e., the model parameters were
Altruistic acceptance of risk 7.1 (2.6) estimated based on the training data), it could be used to predict
Risk factor the label of a new test example (i.e., scores of PTSD/depression
Stress due to social isolation 7.6 (2.3) scale). The output of the model is the predicted clinical score
Psychiatric symptoms obtained during the test phase. The sum of each psychometric
Model1 (PTSD, PCL-5) 28.6 (17.7) question score was included in the model separately: professional
Model2 (Depression, PHQ-9) 10.7 (6.8) recognition before the pandemic (1) and during the pandemic
(2), altruistic acceptance of risk (3) and the level of stress due to
being isolated from one’s family (4).
Linear epsilon-insensitive support vector machine (ε-SVM)
confirmation that one might have transmitted the virus someone regression was applied to predict posttraumatic symptoms and
very close (coworker, partner, friend or family). All of these depression symptoms based on psychometric questions. The
items are in accordance with criteria A for the development choice of machine learning algorithm depends on many factors,
of PTSD in the DSM-5. A trauma index question was also such as the generalization performance measured on test data
used that asked participants to choose their worst experience and the computational cost of the algorithm. In this study, we
considering the previous questions and how long ago the applied a non-kernel regression algorithm: the linear ε-SVM.
event occurred (less or more than 1 month ago). After In preliminary investigations, we compared the performance
completing this questionnaire, the participant indicated the worst of three different algorithms currently available in PRoNTo: ε-
trauma experienced. SVM, gaussian process regression (GPR) (50) and kernel ridge
regression (KRR) (51). There were no significant differences
Depression in performance among the three different approaches. For
The Patient Health Questionnaire 9 (PHQ-9) is a 9-item self- the sake of brevity, we chose to present results only for ε-
report questionnaire that assesses symptoms of major depression SVM. Furthermore, SVM is considered better than most of
based on the DSM-IV criteria (47). The nine symptoms are the other algorithms used because it has better accuracy in its
depressed mood, anhedonia, problems with sleep, tiredness or results, especially for smaller samples. Since their introduction
lack of energy, change in appetite or weight, feelings of guilt in 1992, SVMs have been studied, generalized, and applied to
or worthlessness, problems with concentration, feeling slow or several problems. Furthermore, SVM is relatively stable and
restless and thoughts of suicide. The PHQ-9 score ranges from 0 memory efficient and has been extensively used for regression
to 27 points, and each of the 9 questions can be scored from 0 models (52–54).
(not at all) to 3 (nearly every day). Here, we used the Brazilian– Essentially, ε-SVM performs linear regression in a high-
Portuguese version of the PHQ-9 (48). dimensional space using epsilon-insensitive loss, also known
The table below shows the means and standard deviations as L1 loss. In ε-SVM, the user must set two hyperparameters,
for the psychometric questions and scales (Table 2). Importantly, ε and C, either manually or using a cross-validation scheme.
the level of professional recognition was significantly higher The hyperparameter ε defines a margin of width ε around
during the pandemic than before the COVID-19 pandemic (t- the regression line, setting a margin of “tolerance,” where
test, P < 0.001). any data point that falls within it carries no penalty. The
C hyperparameter, in contrast, controls how strongly data
Pattern Regression Analysis points beyond the epsilon-insensitive margin are penalized.
We used pattern regression analysis to predict mental health In essence, ε sets a margin outside of which data points
outcomes (depression or posttraumatic stress symptoms) based are penalized, and C defines the penalty itself. The idea is
on psychometric questions, including: (1) level of stress due to similar to the concept of a “soft margin” in SVM classification
being isolated from one’s family; (2) professional recognition (55). Both hyperparameters were automatically optimized in
before and during the COVID-19 pandemic; and (3) altruistic PRoNTo using a two-fold nested cross-validation procedure,
acceptance of risk before the pandemic. More specifically, we with the same cross-validation scheme for the internal and
trained two regression models with the goal of predicting external loops.
posttraumatic stress symptoms (model 1) and depression In this case, there are two loops in the cross-validation scheme.
symptoms (model 2). The inner loop is used for parameter optimization, and the
Pattern regression analyses were implemented in the Pattern outer loop is used for assessing the model’s performance. More
Recognition for Neuroimaging Toolbox (PRoNTo), version 3 specifically, the data are divided into training and testing sets
(49). The procedure for building pattern regression models according to the cross-validation scheme selected (outer loop).
consists of two phases: training and testing. During the For each fold of the outer loop, the training set is further divided

Frontiers in Psychiatry | www.frontiersin.org 5 January 2022 | Volume 12 | Article 752870

Portugal et al. PTSD and Depression During Covid-19

FIGURE 1 | Regression models: (a) The training data for the ε-SVM regression model consists of examples that pair the psychometric factors (stress due to social
isolation, altruistic acceptance of risk and professional recognition before and during the pandemic) of each subject and the corresponding clinical score (PCL-5 or
PHQ-9). (b) During the training, the ε-SVM model learns the contribution of each psychometric question for the predictive function. (c) During the testing phase, given
the psychometric questions of a test subject, the ε-SVM model predicts its corresponding clinical score. (d) The model performance is evaluated using three metrics
that measure the agreement between the predicted and actual clinical scores: Pearson’s correlation coefficient (r), coefficient of determination (r2 ) and normalized
mean squared error (NMSE).

into training and testing sets according to the cross-validation transformed into a grid. The parameters used were values of 0.01,
scheme selected (inner/nested loop). The inner loop is used to 0.1, 1, 10, 100, and 1,000.
train and test the model with each value of the hyperparameter To evaluate the ε-SVM performance we used two different
specified by the user. The parameter leading to the highest cross-validation strategies (a two-fold cross-validation and a five-
performance in the inner/nested loop (according to the mean fold cross-validation) to demonstrate that the results were not
squared error) is then used in the outer loop. For each fold dependent on a specific cross-validation scheme. We choose
of the outer loop, the model is trained using the “optimal” two and five-fold cross validation, as these numbers of splits
value of the hyperparameter and tested on the data that were seemed reasonable considering our sample size. A two-fold cross-
omitted (and which were not used for parameter optimization). validation procedure means that the sample was divided in two,
PRoNTo allows for the automatic optimization of more than with half of the sample used for training and half used for testing
one parameter, entered as a cell array of values that is then in the first fold and with the half of the sample that was used

Frontiers in Psychiatry | www.frontiersin.org 6 January 2022 | Volume 12 | Article 752870

Portugal et al. PTSD and Depression During Covid-19

for testing then being used for training and vice versa in the TABLE 3 | Measurements of agreement between the actual and decoded scores
second fold (see Figure 1). The five-fold cross validation involves based on scores of professional recognition, altruistic acceptance of risk and
stress level due to social isolation.
dividing the data into five disjoint sets. Data from each set is left
out once for test and data from the remaining four sets are used Models Cross-validation schemes Measures of agreement
to train the model. This procedure is then repeated five times, so 2
r (P-value) r (P-value) NMSE (P-value)
that each set is left out once. In both cases, the performance of the
model is computed based on the concatenation of the predictions PTSD “Two-fold” 0.35 (0.001) 0.12 (0.001) 0.96 (0.001)
across folds, as implemented in PRoNTo. “Five-fold” 0.34 (0.001) 0.12 (0.001) 0.90 (0.001)
Regarding potential confounders, being female, being younger Depression “Two-fold” 0.36 (0.001) 0.13 (0.001) 0.90 (0.001)
and reporting a current mental health diagnosis have previously “Five-fold” 0.38 (0.001) 0.15 (0.001) 0.86 (0.001)
been associated with depression among essential workers in
* For
Brazil (16). However, removing confounders associated with the reference: corrected p-value = 0.0125.

variable to be predicted (i.e., the labels) is not recommended

because this adjustment is likely to remove not only the variability
in the data due to confounding factors but also the variability specific inferences as in classical (univariate) techniques. Since
in the data associated with the labels (56, 57). To address this each cross-validation fold yields a different weight vector, the
limitation, we balanced the proportion of data from potential final psychometric weight is the average across the folds divided
confounders across the different folds. There was no difference in by its Euclidean norm. For the sake of brevity, we illustrate only
the distribution of the sample regarding the presence of mental the two-fold cross- validation in the manuscript.
disorders diagnosed before the pandemic, gender, age or the
scores on the questions and PTSD/depression symptoms for both RESULTS
cross-validation strategies (see Supplementary Tables 1, 2).
Pattern Regression Model
Performance of the Model After correction for multiple comparisons (since four different
To determine the performance of the regression model, three models were tested, the significance threshold was 0.05/4 =
metrics were used to measure the agreement between the 0.0125), the ε-SVM regression models significantly predicted
predicted and actual PTSD/depression symptoms: Pearson’s PTSD and depression symptoms from the psychometric
correlation coefficient (r), the coefficient of determination (r2 ) questions that potentially represented vulnerability/protective
and the normalized mean squared error (NMSE). The correlation factors for mental disorders. For PTSD, the performance of
coefficient (r) describes the strength of a linear relationship the regression model is presented in Table 3 [twofold: r =
between two variables. A small correlation is an indication of 0.35 (P-value = 0.001), r2 = 0.12 (P-value = 0.001) and
poor predictive performance. The coefficient of determination NMSE = 0.96 (P-value = 0.001), and five-fold: r = 0.34
(r2 ) can be interpreted as the proportion of variance explained by (P-value = 0.001), r2 = 0.12 (P-value = 0.001) and NMSE
the regression. The NMSE is the mean of the squared differences = 0.90 (P-value = 0.001)]. Figure 2A shows a scatter plot
between the predicted and true scores; it represents the mean depicting the predicted vs. actual PTSD symptoms for the
error between the predicted and actual scores and is commonly two-fold cross-validation. Similar results were obtained for
used to evaluate the performance of predictive models. The MSE depression symptoms [two-fold: r = 0.36 (P-value = 0.001),
was normalized by dividing the MSE by the variance in the r2 = 0.13 (P-value = 0.001) and NMSE = 0.90 (P-value =
target values. 0.001), and five-fold: r = 0.38 (P-value = 0.001), r2 = 0.15
The significance of the regression performance measures (P-value = 0.001) and NMSE = 0.86 (P-value = 0.001)],
was determined using permutation tests, i.e., the same cross- indicating that our models significantly decoded both PTSD
validation procedure described above was performed 1,000 times and depression symptoms from psychometric questions (Table 3;
with the labels permuted across the participants. The P-value was Figure 2B). There were no significant differences in performance
calculated by counting how many times the absolute value of the among the different kernel regression approaches and non-
metric with the permuted labels was equal to or greater (less for kernel approaches (see Supplementary Tables 1, 3 to consistency
MSE) than the absolute value of the metric obtained with the of results).
correct labels and dividing by 1,000. The results were considered
significant when the model performed equal to or better than the Contributions of Psychometric Questions
model without shuffling the labels at most 5% of the time across to the Regression Model
1,000 permutations (58). For the sake of brevity, we display the weight maps only
for the model based on the two-fold cross-validation scheme
Model Interpretation in the main manuscript. The relative contribution of each
The weights represent the contribution of each psychometric psychometric question to the ε-SVM for both models is
question to the linear predictive function and can be explicitly shown in Figure 3. The weight of each psychometric question
computed and plotted for interpretation and discussion. As corresponds to its contribution to the model’s prediction.
previously discussed in the literature (58), the weight map of Notably, for the PTSD model, the psychometric questions
linear machine learning models cannot be thresholded to make with the greatest contributions were the level of stress

Frontiers in Psychiatry | www.frontiersin.org 7 January 2022 | Volume 12 | Article 752870

Portugal et al. PTSD and Depression During Covid-19

FIGURE 2 | Scatter plots of actual vs. predicted values applying a two-fold cross-validation scheme for the PTSD symptoms model and for the depression model. (A)
Scatter plot between the actual and predicted PCL-5 scores (PTSD symptoms model). (B) Scatter plot between the actual and predicted PHQ-9 scores
(depression model).

due to social isolation (0.85) and professional recognition, have higher levels of self-perceived levels of stress due to
mainly before the pandemic (before = −0.49 and during = being isolated from one or more members of their families
−0.18), and the psychometric question making the smallest could be more vulnerable to experiencing psychiatric symptoms.
contribution was altruistic acceptance of risk (0.03). Similar Furthermore, our results indicate that professional recognition
results were obtained for the depression model, in which the might be an important protective factor for the mental health
level of stress due to social isolation (0.85) and professional of hospitals and emergency care workers. Finally, our results are
recognition (before = −0.49 and during = −0.18) made promising since they suggest that machine learning algorithms
the greatest contributions and the psychometric question could provide significant models for predicting mental health
making the smallest contribution was altruistic acceptance of symptoms from psychometric data. To our knowledge, this study
risk (0.03). Interestingly, professional recognition had negative is the first showing that the perception of stress from being
predictive value, indicating inverse relationships with PTSD isolated and professional recognition are very important factors
and depression. to be considered for Brazilian healthcare workers’ mental health
conditions. Such knowledge is relevant for devising preventive
DISCUSSION measures and care actions at occupational and institutional
levels, considering the importance of the current context.
There were many new cases and deaths during the data collection An important strength of the present study was the assessment
(3), revealing the pandemic’s impact in Brazil. Consequently, of traumatic events specifically related to COVID-19 pandemics.
intense demand was imposed on healthcare workers, leading The participants answered questions that investigated potentially
to greater pressure on mental health services in Brazil. The traumatic situations experienced by healthcare workers since
main goal of the present study was to apply machine learning, the COVID-19 outbreak. The traumatic experience related to
particularly pattern regression analysis, to determine the impact COVID-19 most frequently reported was “learning about the
of the self-perceived level of stress due to social isolation, death of a close relative or coworker, due to COVID-19” followed
professional recognition and altruistic acceptance of risk on the by “possibly transmitting the COVID-19 virus to another
mental health outcomes (depression and PTSD symptoms) of person.” This finding is in agreement with previous studies about
employees working in hospitals and/or emergency care services trauma prevalence before the pandemic since trauma related to
during the COVID-19 pandemic. The results confirmed that death of a beloved one has been reported to be the most frequent
ε-SVM models were able to predict PTSD symptoms (PCL- trauma (59).
5 scores) and depression symptoms (PHQ-9 scores). For both Throughout this pandemic healthcare workers have had to
models, the self-perceived level of stress due to social isolation self-isolate from their own families mainly due to fear of
and professional recognition were the variables making the transmitting the virus to their loved ones. However, since humans
greatest contributions to the predictive function. Interestingly, are highly social and cooperative animals, the response to the
professional recognition had negative predictive value, indicating threat of infection by COVID-19 causes the desire for physical
an inverse relationship with posttraumatic and depression contact, especially in relation to loved ones, such as family
symptoms. These findings suggest that hospital workers who members (60). In fact, adequate social contact is critical for

Frontiers in Psychiatry | www.frontiersin.org 8 January 2022 | Volume 12 | Article 752870

Portugal et al. PTSD and Depression During Covid-19

FIGURE 3 | (A) Plot showing the values of the weights for each scale for the prediction of PTSD symptoms. (B) Plot showing the values of the weights for each scale
for the prediction of depression symptoms.

mental health (61). For these reasons, humans struggle when relevant factor for PTSD and depression symptoms. This finding
forced to live in isolation, and most of us find social deprivation is supported by the existing literature on previous epidemics
stressful. In fact, our data indicate that the loss of this type and the current COVID-19 pandemic that has reported negative
of social contact can impact the mental health of healthcare associations between quarantine/social isolation and mental
workers. In our pattern regression models, the level of stress health outcomes in these professionals (11, 12, 62, 63). Along
due to isolation from one’s family for at least one week was a the same line, a recent meta-analysis focusing on objective

Frontiers in Psychiatry | www.frontiersin.org 9 January 2022 | Volume 12 | Article 752870

Portugal et al. PTSD and Depression During Covid-19

measures of isolation reported that individuals experiencing outbreak. In contrast, other studies have reported that altruistic
isolation or quarantine were at increased risk for adverse mental acceptance of risk was not related to psychiatric symptoms
health outcomes, particularly after a duration of 1 week or among hospital employees following epidemic outbreaks and the
longer (29). To the best of our knowledge, our study is the general population (75, 76). However, as emphasized above, these
first to show that the self-perceived level of stress due to being findings should be interpreted with caution since all predictive
isolated from one’s family members is a significant and important questions contributed to the final prediction.
factor for the severity of psychiatric symptoms in hospital and Machine learning tools, specifically pattern regression, have
emergency care workers during the COVID-19 pandemic. One been successfully applied with many types of data, such as
possible explanation for why self-perception of stress leads to neuroimaging data (33–35, 77, 78). However, their use has
psychiatric symptoms came from a study showing that social been less investigated in studies using psychometric data (79).
isolation (self-perception of loneliness during COVID-19) both In the context of the COVID-19 pandemic, few studies have
mediates and moderates the indirect effect of COVID-19 worries applied pattern regression based on psychometric data to predict
on posttraumatic stress symptoms (PTSS) related to COVID-19 continuous variables among general samples (36, 37) and
among individuals who have not yet been infected with COVID- university student samples (38). Our results are promising since
19 (64). Further studies should investigate the relationship they suggest that machine learning algorithms could provide
among COVID-19-related worries, feeling of loneliness and the significant models for predicting mental health symptoms from
self-perceived level of stress due to being isolated in predicting psychometric data.
PTSD and depression symptoms in healthcare workers. There were also some limitations to the present study.
Conversely, we found that self-perceived professional First, the sample was not representative of the entire Brazilian
recognition before and during the pandemic had negative healthcare worker population since the data were obtained by
predictive value, indicating an inverse relationship with PTSD a convenience snowball sampling technique via a link sent by
and depression symptoms. Importantly, the level of perceived WhatsApp and e-mail. While online recruitment guarantees
professional recognition was higher during the COVID-19 large samples, it does not guarantee sample representativeness.
pandemic than before it. Here, professional recognition refers to To reduce this limitation, we contacted all major healthcare
the recognition of a person’s work by the general population and worker groups in Brazil to publish the main project proposal
reflects the following factors: (1) the esteem support factor, which and the link to complete the survey online on their websites
is a type of social support that reassures a person about his or her and on Instagram. Additionally, there might have been selection
skills (65); and (2) the construction of the social esteem factor, bias. For example, the southeastern region (73.5%) was
which is a sense of the recognition of a person’s achievements and overrepresented, and we cannot ignore that our results could
contributions at work (66). In fact, both factors are negatively have been driven by the highest socioeconomic region in Brazil.
associated with negative mental health problems, including For example, death and comorbid disease were more common
burnout symptoms (67–69). We believe that one of the pathways among Brazilians from the North region than among those
by which professional recognition might protect against the from the Central-South regions (80). Furthermore, the worst
severity of posttraumatic and depression symptoms is enhancing public health and social scenarios were present in the northern
social support and self-esteem among these professionals. regions of Brazil (81). These regions were underrepresented in
Furthermore, professional recognition has been shown to our sample (17.16%), and it seems important to emphasize that
enhance self-determination (70) and work satisfaction (71). this scenario seen in the Northeast/North regions could worsen
In the current pandemic, the findings regarding professional the consequences of COVID-19 on the mental health of health
recognition have shown that the recognition of their work and care workers. Second, the use of self-report measures did not
efforts by hospital management could be motivating factors for enable us to verify the reliability of the responses or to ensure that
medical staff to continue working effectively (72). Our findings participants correctly understood the questions. Furthermore, to
are in line with Barello’s (32) results and extend prior findings minimize that we did not apply any objective quality control to
to other psychiatric conditions, showing that professional ensure that the online survey results were credible, we offered
recognition might be considered a relevant protective factor anonymity on self-administered questionnaires to reduce social
for the severity of posttraumatic and depression symptoms in desirability bias, and we also attempted to develop a more concise
healthcare workers during the COVID-19 pandemic. questionnaire to avoid tiredness. Future research should seek to
Finally, the psychometric question with the lowest compare the present study data with those collected using other
contribution to the model’s predictive function was altruistic methods (e.g., semistructured interviews, qualitative approaches,
acceptance of risk, a quality frequently found among healthcare etc.). Another important limitation is that, since removing
workers (73, 74). One possible explanation for our findings confounders associated with the variable to be predicted is not
is that the role of altruistic acceptance of risk as a buffer recommended, we cannot exclude our results perhaps being
against psychiatric symptoms is inconsistent. Some studies have influenced by some objective aspects that impact the mental
found that altruistic acceptance of risk was negatively related health of healthcare workers during the COVID-19 pandemic,
to psychiatric symptoms in healthcare workers following an such as working years, professional level, and working on the
epidemic outbreak (11, 12). These findings indicate that altruistic front line of the hospital. Furthermore, although we used two
acceptance of risk might have protected some hospital employees cross-validation schemes (two-fold (or half split) cross-validation
against negative psychological outcomes following the epidemic and five-fold cross-validation), predictive models should ideally

Frontiers in Psychiatry | www.frontiersin.org 10 January 2022 | Volume 12 | Article 752870

Portugal et al. PTSD and Depression During Covid-19

be further validated with a truly independent sample. Finally, for example, creating a program to improve the self-perception
with regard to hyperparameter optimizations, it should be noted of being recognized by the institution can be a very effective
that a more automatic fine-tuning technology such as Bayesian way to protect the mental health of these professionals.
optimization may be a good option in the future (82, 83). Within this framework of thought, practical issues, such as
The COVID-19 pandemic is still unfolding, and it is likely salary valorization and improvement of work environment
that the virus and its consequences will impact the health conditions, which include work healthcare centers, readjustment
system for some time to come. Identifying vulnerability and of work environments, humanized leaderships, suppliers of
protective factors to prevent mental disorders from progressing consumables, materials, and individual protection equipment,
in healthcare professionals is necessary to promote prevention could contribute to the perception of professional recognition
strategies and to counteract stressors and challenges during this (92). Additionally, outreach in the media and government
outbreak. Our study findings draw our attention to the predictive could encourage the population to recognize issues concerning
role of the level of self-perceived stress due to social isolation in professional significance. In addition, technical support
the severity of PTSD and depression symptoms. Furthermore, (room for communication, support staffed by mental health
our findings emphasize the protective role of professional professionals, periodic monitoring of mental health, space to
recognition in posttraumatic and depression symptomatology. therapeutic interventions, psychoeducation about symptoms
We suggest here that self-perceived stress due to social distancing for early identification of mental disorders, meditation and
and self-perceived professional recognition might also represent mindfulness techniques, physical exercise incentives, etc.) could
important vulnerabilities to be assessed in clinical interviews. demonstrate that professionals’ mental health is appreciated
Bringing these aspects into the clinical assessment could help by the organization (93). Finally, we hope that the COVID-19
clinicians to estimate the risk of worsening PTSD or depression pandemic will prompt the recognition of the contributions
in these professionals. Based on our findings, appropriate action of all healthcare workers with appropriate protection and
to monitor and reduce the level of stress due to social isolation compensation, such as wage appreciation.
from family among these groups of individuals working on the In summary, this study showed that a machine learning
frontline of the pandemic should be undertaken immediately. approach (pattern regression model) was able to predict mental
Measuring the degree of self-perceived stress due to social health outcomes and PTSD and depression symptoms in
isolation is an important addition to mental health assessments healthcare workers based on the self-perceived level of stress
during the COVID-19 pandemic. Stress and social isolation can due to isolation and professional recognition. These results
impact health and immune function, for example, decreasing add to the literature indicating the importance of considering
inflammatory control and viral immunity (84–86). Therefore, how each healthcare worker perceives the stress of isolation
reducing the level of stress due to social isolation is essential and professional recognition, in addition to more objective
during a time when individuals require strong immune function factors, such as years of work, professional level, gender,
to fight off a novel virus. For instance, one possible action to and age. We suggest that it is a fundamental aspect of
mitigate the consequences of the level of stress due to being implementing targeted clinical evaluations and intervention
isolated is to encourage vulnerable individuals to remain in programs within institutions to reduce the psychiatric burden on
regular contact with family and friends through video chats, health systems worldwide during the COVID-19 pandemic and
phone calls and online groups. The use of video-embedded even future pandemics.
digital communication is likely to gain importance. The visual
component of interpersonal encounters appears to play a key
role in creating a more satisfying experience of digital social
DATA AVAILABILITY STATEMENT
media (87). Strategies to foster a sense of belonging among The raw data supporting the conclusions of this article will be
healthcare workers should be encouraged. For example, being made available by the authors, without undue reservation.
connected with or reading stories from people who are also
isolated from their families can promote identification and,
consequently, emotional comfort. In fact, sense of belonging ETHICS STATEMENT
is a key buffering factor against feelings of stress among
healthcare workers during the COVID-19 pandemic (88). Work The studies involving human participants were reviewed and
environments that facilitate these basic psychological needs to approved by Research Ethics Committee of the Faculty of
feel connected to others and to have a sense of belonging prompt Medicine, UFF. The patients/participants provided their written
positive psychological outcomes, such as enhanced performance informed consent to participate in this study.
and greater psychological well-being (89–91).
This mobilization now will allow the public health system AUTHOR CONTRIBUTIONS
to apply the knowledge gained to any future periods of
increased infection and lockdowns, which will be particularly LP analyzed and interpreted the data, wrote the manuscript,
crucial for healthcare workers and to future pandemics. Our and contributed to the data organization. CG contributed to
findings strongly suggest that positive recognition experiences the data organization, data interpretation, and revision of the
can be fostered by hospital management to buffer against manuscript. RG, FE, IM, KT, MM, EV, and ID interpreted data
negative effects on mental health among healthcare workers and revised the manuscript. MP contributed to writing and

Frontiers in Psychiatry | www.frontiersin.org 11 January 2022 | Volume 12 | Article 752870

Portugal et al. PTSD and Depression During Covid-19

revising the manuscript and the data interpretation. LO analyzed ACKNOWLEDGMENTS

and interpreted the data and wrote and revised the manuscript.
All of the authors read and approved the final manuscript. We thank Dr. Ivan for his valuable help with the
data interpretation.
FUNDING
SUPPLEMENTARY MATERIAL
This work (data collection, analysis and writing) was supported
in part by federal and state Brazilian research agencies (CNPq The Supplementary Material for this article can be found
and FAPERJ). Scholarships were awarded by the federal Brazilian online at: https://www.frontiersin.org/articles/10.3389/fpsyt.
research agency CAPES 614 001, CAPES/PRINT. 2021.752870/full#supplementary-material

REFERENCES 15. Holmes EA, O’Connor RC, Perry VH, Tracey I, Wessely S, Arseneault L,
et al. Multidisciplinary research priorities for the COVID-19 pandemic: a
1. World Health Organization. WHO Characterizes COVID-19 as a Pandemic. call for action for mental health science. Lancet Psychiatry. (2020) 7:547–
(2020). Available online at: https://www.who.int/emergencies/diseases/novel- 60. doi: 10.1016/S2215-0366(20)30168-1
coronavirus-2019/events-as-they-happen (accessed July 23, 2020). 16. De Boni RB, Balanzá-Martínez V, Mota JC, Cardoso TA, Ballester P, Atienza-
2. World Health Organization. Coronavirus Disease (COVID-2019) Situation Carbonell B, et al. Depression, anxiety, and lifestyle among essential workers:
Reports. (2021). Available online at: https://www.who.int/publications/m/ a web survey from Brazil and Spain during the COVID-19 pandemic. J Med
item/weekly-epidemiological-update_on-covid-19 (accessed April 20, 2021). Internet Res. (2020) 22:e22835. doi: 10.2196/22835
3. Worldometer. COVID-19 Coronavirus Pandemic. World / Countries / Brazil 17. Hu Z, Chen B. The status of psychological issues among frontline health
(2020). Avaialble online at: https://www.worldometers.info/coronavirus/ workers confronting the coronavirus disease 2019 pandemic. Front Public
country/brazil/#graph-cases-daily (accessed July 9, 2021). Health. (2020) 8:265. doi: 10.3389/fpubh.2020.00265
4. The Lancet. COVID-19 in Brazil: “so what?”. Lancet. (2020) 18. Country Profiles-Brazil, Spain. Seattle, WA: IHME, University of
395:1461. doi: 10.1016/S0140-6736(20)31095-3 Washington (2018).
5. Xiang YT, Yang Y, Li W, Zhang L, Zhang Q, Cheung T, et al. Timely mental 19. Gupta S, Sahoo S. Pandemic and mental health of the front-line
health care for the 2019 novel coronavirus outbreak is urgently needed. Lancet healthcare workers: a review and implications in the Indian context amidst
Psychiatry. (2020) 7:228–9. doi: 10.1016/S2215-0366(20)30046-8 COVID-19. Gen Psychiatr. (2020) 33:e100284. doi: 10.1136/gpsych-2020-1
6. Varghese A, George G, Kondaguli SV, Naser AY, Khakha DC, Chatterji 00284
R. Decline in the mental health of nurses across the globe during 20. Hao Q, Wang D, Xie M, Tang Y, Dou Y, Zhu L, et al. Prevalence and risk factors
COVID-19: a systematic review and meta-analysis. J Glob Health. (2021) of mental health problems among healthcare workers during the COVID-
11:05009. doi: 10.7189/jogh.11.05009 19 pandemic: a systematic review and meta-analysis. Front Psychiatry. (2021)
7. Cénat JM, Blais-Rochette C, Kokou-Kpolou CK, Noorishad PG, Mukunzi JN, 12:567381. doi: 10.3389/fpsyt.2021.567381
McIntee SE, et al. Prevalence of symptoms of depression, anxiety, insomnia, 21. Shaukat N, Ali DM, Razzak J. Physical and mental health impacts of COVID-
posttraumatic stress disorder, and psychological distress among populations 19 on healthcare workers: a scoping review. Int J Emerg Med. (2020)
affected by the COVID-19 pandemic: a systematic review and meta-analysis. 13:40. doi: 10.1186/s12245-020-00299-5
Psychiatry Res. (2021) 295:113599. doi: 10.1016/j.psychres.2020.113599 22. Lu W, Wang H, Lin Y, Li L. Psychological status of medical workforce during
8. Shreffler J, Petrey J, Huecker M. The impact of COVID-19 on healthcare the COVID-19 pandemic: a cross-sectional study. Psychiatry Res. (2020)
worker wellness: a scoping review. West J Emerg Med. (2020) 21:1059– 288:112936. doi: 10.1016/j.psychres.2020.112936
66. doi: 10.5811/westjem.2020.7.48684 23. Roma P, Monaro M, Colasanti M, Ricci E, Biondi S, Di Domenico A, et al.
9. Iqbal MR, Chaudhuri A. COVID-19: results of a national survey A 2-month follow-up study of psychological distress among Italian people
of United Kingdom healthcare professionals’ perceptions of current during the COVID-19 lockdown. Int J Environ Res Public Health. (2020)
management strategy - a cross-sectional questionnaire study. Int J Surg. (2020) 17:8180. doi: 10.3390/ijerph17218180
79:156–61. doi: 10.1016/j.ijsu.2020.05.042 24. d’Ettorre G, Ceccarelli G, Santinelli L, Vassalini P, Innocenti GP, Alessandri F,
10. Galanis P, Vraka I, Fragkou D, Bilali A, Kaitelidou D. Seroprevalence et al. Post-traumatic stress symptoms in healthcare workers dealing with the
of SARS-CoV-2 antibodies and associated factors in healthcare workers: COVID-19 pandemic: a systematic review. Int J Environ Res Public Health.
a systematic review and meta-analysis. J Hosp Infect. (2021) 108:120– (2021) 18:601. doi: 10.3390/ijerph18020601
34. doi: 10.1016/j.jhin.2020.11.008 25. Xiao X, Zhu X, Fu S, Hu Y, Li X, Xiao J. Psychological impact of
11. Wu P, Fang Y, Guan Z, Fan B, Kong J, Yao Z, et al. The psychological healthcare workers in China during COVID-19 pneumonia epidemic: a multi-
impact of the SARS epidemic on hospital employees in China: exposure, center cross-sectional survey investigation. J Affect Disord. (2020) 274:405–
risk perception, and altruistic acceptance of risk. Can J Psychiatry. (2009) 10. doi: 10.1016/j.jad.2020.05.081
54:302–11. doi: 10.1177/070674370905400504 26. Pancani L, Marinucci M, Aureli N, Riva P. Forced social isolation and mental
12. Liu X, Kakade M, Fuller CJ, Fan B, Fang Y, Kong J, et al. Depression health: a study on 1,006 Italians under COVID-19 lockdown. Front Psychol.
after exposure to stressful events: lessons learned from the severe (2021) 12:663799. doi: 10.3389/fpsyg.2021.663799
acute respiratory syndrome epidemic. Compr Psychiatry. (2012) 53:15– 27. Giardino DL, Huck-Iriart C, Riddick M, Garay A. The endless quarantine:
23. doi: 10.1016/j.comppsych.2011.02.003 the impact of the COVID-19 outbreak on healthcare workers after three
13. Marvaldi M, Mallet J, Dubertret C, Moro MR, Guessoum months of mandatory social isolation in Argentina. Sleep Med. (2020) 76:16–
SB. Anxiety, depression, trauma-related, and sleep disorders 25. doi: 10.1016/j.sleep.2020.09.022
among healthcare workers during the COVID-19 pandemic: a 28. Gonçalves AP, Zuanazzi AC, Salvador AP, Jaloto A, Pianowski G, Carvalho LF.
systematic review and meta-analysis. Neurosci Biobehav Rev. (2021) Preliminary findings on the associations between mental health indicators and
126:252–64. doi: 10.1016/j.neubiorev.2021.03.024 social isolation during the COVID-19 pandemic. Arch. Psychiatry Psychother.
14. Lai J, Ma S, Wang Y, Cai Z, Hu J, Wei N, et al. Factors (2020) 22:10–9. doi: 10.12740/APP/122576
associated with mental health outcomes among health care workers 29. Henssler J, Stock F, van Bohemen J, Walter H, Heinz A, Brandt L.
exposed to coronavirus disease 2019. JAMA Netw Open. (2020) Mental health effects of infection containment strategies: quarantine
3:e203976. doi: 10.1001/jamanetworkopen.2020.3976 and isolation-a systematic review and meta-analysis. Eur Arch

Frontiers in Psychiatry | www.frontiersin.org 12 January 2022 | Volume 12 | Article 752870

Portugal et al. PTSD and Depression During Covid-19

Psychiatry Clin Neurosci. (2021) 271:223–34. doi: 10.1007/s00406-020-0 for the Brazilian context. Trends Psychiatry Psychother. (2016) 38:207–
1196-x 15. doi: 10.1590/2237-6089-2015-0074
30. Hwang TJ, Rabheru K, Peisah C, Reichman W, Ikeda M. Loneliness and 47. Kroenke K, Spitzer RL, Williams JB. The PHQ-9: validity of
social isolation during the COVID-19 pandemic. Int Psychogeriatr. (2020) a brief depression severity measure. J Gen Intern Med. (2001)
32:1217–20. doi: 10.1017/S1041610220000988 16:606–13. doi: 10.1046/j.1525-1497.2001.016009606.x
31. Khalid I, Khalid TJ, Qabajah MR, Barnard AG, Qushmaq IA. Healthcare 48. de Lima Osório F, Vilela Mendes A, Crippa JA, Loureiro SR. Study
workers emotions, perceived stressors and coping strategies during a MERS- of the discriminative validity of the PHQ-9 and PHQ-2 in a sample
CoV outbreak. Clin Med Res. (2016) 14:7–14. doi: 10.3121/cmr.2016.1303 of Brazilian women in the context of primary health care. Perspect
32. Barello S, Palamenghi L, Graffigna G. Burnout and somatic Psychiatr Care. (2009) 45:216–27. doi: 10.1111/j.1744-6163.2009.
symptoms among frontline healthcare professionals at the peak 00224.x
of the Italian COVID-19 pandemic. Psychiatry Res. (2020) 49. Schrouff J, Mourão-Miranda J, Phillips C, Parvizi J. Decoding intracranial
290:113129. doi: 10.1016/j.psychres.2020.113129 EEG data with multiple kernel learning method. J Neurosci Methods. (2016)
33. Portugal LC, Rosa MJ, Rao A, Bebko G, Bertocci MA, Hinze 261:19–28. doi: 10.1016/j.jneumeth.2015.11.028
AK, et al. Can emotional and behavioral dysregulation in youth 50. Rasmussen CE, Williams CKI. Gaussian processes for machine learning. Int J
be decoded from functional neuroimaging?. PLoS ONE. (2016) Neural Syst. (2006) 14:69–106. doi: 10.1142/S0129065704001899
11:e0117603. doi: 10.1371/journal.pone.0117603 51. Hoerl AE, Kennard RW. Ridge regression: biased estimation
34. Oliveira L, Portugal LCL, Pereira M, Chase HW, Bertocci M, Stiffler for nonorthogonal problems. Technometrics. (1970) 42:80–
R, et al. Predicting bipolar disorder risk factors in distressed young 6. doi: 10.1080/00401706.2000.10485983
adults from patterns of brain activation to reward: a machine learning 52. Rivas-Perea P, Cota-Ruiz J, Chaparro DG, Venzor JAP, Carreón AQ, Rosiles
approach. Biol Psychiatry Cogn Neurosci Neuroimaging. (2019) 4:726– JG. Support vector machines for regression: a succinct review of large-
33. doi: 10.1016/j.bpsc.2019.04.005 scale and linear programming formulations. Int J Intell Sci. (2013) 3:5–
35. Portugal LCL, Schrouff J, Stiffler R, Bertocci M, Bebko G, Chase 14. doi: 10.4236/ijis.2013.31002
H, et al. Predicting anxiety from wholebrain activity patterns to 53. Clarke SM, Griebsch JH, Simpson TW. Analysis of support vector regression
emotional faces in young adults: a machine learning approach. for approximation of complex engineering analyses. ASME. J Mech Des.
Neuroimage Clin. (2019) 23:101813. doi: 10.1016/j.nicl.2019. (2005) 127:1077–87. doi: 10.1115/1.1897403
101813 54. Zhang F, O’Donnell LJ. Chapter 7: support vector regression. In:
36. Prout TA, Zilcha-Mano S, Aafjes-van Doorn K, Békés V, Christman- Andrea Mechelli A, Vieira S, editors. Machine Learning: Methods
Cohen I, Whistler K, et al. Identifying predictors of psychological distress and Applications to Brain Disorders. Elsevier (2020). p. 123–40.
during COVID-19: a machine learning approach. Front Psychol. (2020) doi: 10.1016/B978-0-12-815739-8.00007-9
11:586202. doi: 10.3389/fpsyg.2020.586202 55. Vergun S, Deshpande AS, Meier TB, Song J, Tudorascu DL, Nair VA,
37. Jha IP, Awasthi R, Kumar A, Kumar V, Sethi T. Explainable-machine-learning et al. Characterizing functional connectivity differences in aging adults using
to discover drivers and to predict mental illness during COVID-19. medRxiv. machine learning on resting state fMRI data. Front Comput Neurosci. (2013)
(2020). doi: 10.1101/2020.07.19.20157164 7:38. doi: 10.3389/fncom.2013.00038
38. Herbert C, El Bolock A, Abdennadher S. How do you feel during the 56. Rao A, Monteiro JM, Mourao-Miranda J. Alzheimer’s disease initiative.
COVID-19 pandemic? A survey using psychological and linguistic predictive modelling using neuroimaging data in the presence of confounds.
self-report measures, and machine learning to investigate mental Neuroimage. (2017) 150:23–49. doi: 10.1016/j.neuroimage.2017.01.066
health, subjective experience, personality, and behaviour during the 57. Miller GA, Chapman JP. Misunderstanding analysis of covariance. J Abnorm
COVID-19 pandemic among university students. BMC Psychol. (2021) Psychol. (2001) 110:40–8. doi: 10.1037/0021-843X.110.1.40
9:90. doi: 10.1186/s40359-021-00574-x 58. Schrouff J, Monteiro JM, Portugal L, Rosa MJ, Phillips C, Mourão-
39. Hastie T, Tibshirani R, Friedman J. The Elements of Statistical Learning: Miranda J. Embedding anatomical or functional knowledge in whole-
Data Mining, Inference, and Prediction. New York, NY: Springer Publishing brain multiple kernel learning models. Neuroinformatics. (2018) 16:117–
Company, Inc (2001). 43. doi: 10.1007/s12021-017-9347-8
40. Cohen JR, Asarnow RF, Sabb FW, Bilder RM, Bookheimer SY, 59. Luz MP, Coutinho ES, Berger W, Mendlowicz MV, Vilete LM, Mello MF,
Knowlton BJ, et al. Decoding continuous variables from neuroimaging et al. Conditional risk for posttraumatic stress disorder in an epidemiological
data: basic and clinical applications. Front Neurosci. (2011) study of a Brazilian urban population. J Psychiatr Res. (2016) 72:51–
5:75. doi: 10.3389/fnins.2011.00075 7. doi: 10.1016/j.jpsychires.2015.10.011
41. Orrù G, Monaro M, Conversano C, Gemignani A, Sartori G. Machine 60. Dezecache G, Frith CD, Deroy O. Pandemics and the great evolutionary
learning in psychometrics and psychological research. Front Psychol. (2020) mismatch. Curr Biol. (2020) 30:R417–9. doi: 10.1016/j.cub.2020.04.010
10:2970. doi: 10.3389/fpsyg.2019.02970 61. Hawkley LC, Cacioppo JT. Loneliness matters: a theoretical and empirical
42. CONASEMS, Conselho Nacional de Secretarias Municipais de Saúde. review of consequences and mechanisms. Ann Behav Med. (2010) 40:218–
Protagonismo feminino na saúde: mulheres são a maioria nos serviços e na 27. doi: 10.1007/s12160-010-9210-8
gestão do SUS. (2020). Available online at: https://www.conasems.org.br/o- 62. Maunder RG, Lancee WJ, Rourke S, Hunter JJ, Goldbloom D, Balderson
protagonismo-feminino-na-saude-mulheres-sao-a-maioria-nos-servicos-e- K, et al. Factors associated with the psychological impact of severe acute
na-gestao-do-sus/ (accessed March 6, 2020). respiratory syndrome on nurses and other hospital workers in Toronto.
43. Chong MY, Wang WC, Hsieh WC, Lee CY, Chiu NM, Yeh WC, Psychosom Med. (2004) 66:938–42. doi: 10.1097/01.psy.0000145673.84
et al. Psychological impact of severe acute respiratory syndrome on 698.18
health workers in a tertiary hospital. Br J Psychiatry. (2004) 185:127– 63. Urufia WOD, Wahiduddin, Russeng SS, Nadjib Bustan M, Amiruddin R,
33. doi: 10.1192/bjp.185.2.127 Bintara Birawida A. Risk factors of posttraumatic stress symptoms among
44. American Psychiatric Association (APA) Diagnostic and Statistical Manual healthcare workers during Covid-19 pandemic in Regional General Hospital
of Mental Disorders. 5th edition. Washington, DC: American Psychiatric of Baubau City, Southeast Sulawesi. Ann Romani Soc Cell Biol. (2021) 25:727–
Association (2013). 43. Retrieved from: https://www.annalsofrscb.ro/index.php/journal/article/
45. Weathers FW, Litz BT, Keane TM, Palmieri PA, Marx BP, Schnurr PP. PTSD: view/250
National Center for PTSD. The PTSD checklist for DSM-5 (PCL-5). (2013). 64. Boyraz G, Legros DN, Tigershtrom A. COVID-19 and
Available online at: https://www.ptsd.va.gov/professional/assessment/adult- traumatic stress: the role of perceived vulnerability, COVID-19-
sr/ptsd-checklist.asp (accessed March 21, 2019). related worries, and social isolation. J Anxiety Disord. (2020)
46. Lima EP, Vasconcelos AG, Berger W, Kristensen CH, Nascimento ED, 76:102307. doi: 10.1016/j.janxdis.2020.102307
Figueira I, et al. Cross-cultural adaptation of the Posttraumatic Stress 65. Cohen S, Wills TA. Stress, social support, and the buffering hypothesis. Psychol
Disorder Checklist 5 (PCL-5) and Life Events Checklist 5 (LEC-5) Bul. (1985) 98:310–57. doi: 10.1037/0033-2909.98.2.310

Frontiers in Psychiatry | www.frontiersin.org 13 January 2022 | Volume 12 | Article 752870

Portugal et al. PTSD and Depression During Covid-19

66. Honneth A. The Struggle for Recognition: The Moral Grammar of Social 82. Ke H, Ke H, Chen D, Shi B, Zhang J, Liu X, et al. Improving
Conflicts. Cambridge: Oxford; Cambridge, MA: Polity Press; Blackwell (1995). brain e-health services via High917 performance EEG classification with
67. Renger D, Miché M, Casini A. Professional recognition at work: the grouping bayesian optimization. IEEE Trans Serv Comput. (2020) 13:696–
protective role of esteem, respect, and care for burnout among employees. 708. doi: 10.1109/TSC.2019.2962673
J Occup Environ Med. (2020) 62:202–9. doi: 10.1097/JOM.00000000000 83. Wu J, Chen X-Y, Zhang H, Xiong L-D, Lei H, Deng S-H.
01782 Hyperparameter 920 optimization for machine learning models based
68. Bonetti L, Tolotti A, Valcarenghi D, Pedrazzani C, Barello S, Ghizzardi on Bayesian optimization. J Electron Sci Technol. (2019) 17:26–40.
G, et al. Burnout precursors in oncology nurses: a preliminary cross- doi: 10.11989/JEST.1674-862X.80904120
sectional study with a systemic organizational analysis. Sustainability. (2019) 84. Cacioppo S, Cacioppo JT. Decoding the invisible forces of social connections.
11:1246. doi: 10.3390/su11051246 Front Integr Neurosci. (2012) 6:51. doi: 10.3389/fnint.2012.00051
69. Zwack J, Schweitzer J. If every fifth physician is affected by burnout, 85. Pressman SD, Cohen S, Miller GE, Barkin A, Rabin BS, Treanor JJ. Loneliness,
what about the other four? Resilience strategies of experienced social network size, and immune response to influenza vaccination in college
physicians. Acad Med. (2013) 88:382–9. doi: 10.1097/ACM.0b013e31828 freshmen Health Psychol. (2005) 24:297–306. doi: 10.1037/0278-6133.24.3.297
1696b 86. Segerstrom SC, Miller GE. Psychological stress and the human immune
70. Decker C, Van Quaquebeke N. Getting respect from a boss you respect: system: a meta-analytic study of 30 years of inquiry. Psychol Bull. (2004)
how different types of respect interact to explain subordinates’ job 130:601–30. doi: 10.1037/0033-2909.130.4.601
satisfaction as mediated by self-determination. J Bus Ethics. (2015) 131:543– 87. Vlahovic TA, Roberts S, Dunbar R. Effects of duration and laughter on
56. doi: 10.1007/s10551-014-2291-8 subjective happiness within different modes of communication. J Comp Med
71. Clarke N, Mahadi N. The significance of mutual recognition respect in Commun. (2012) 4:436–50. doi: 10.1111/j.1083-6101.2012.01584.x
mediating therelationships between trait emotional intelligence, affective 88. Sangal RB, Wrzesniewski A, Dibenigno J, Reid E, Ulrich A, Liebhardt B, et al.
commitment and job satisfaction. Pers Individ Dif. (2017) 105:129– Work team identification associated with less stress and burnout among front-
34. doi: 10.1016/j.paid.2016.09.028 line emergency department staff amid the COVID-19 pandemic. BMJ Leader.
72. Cai H, Tu B, Ma J, Chen L, Fu L, Jiang Y, et al. Psychological impact and (2020) 5:51–4. doi: 10.1136/leader-2020-000331
coping strategies of frontline medical staff in hunan between January and 89. Baumeister RF, Leary MR. The need to belong: desire for interpersonal
March 2020 during the outbreak of coronavirus disease 2019 (COVID-19) attachments as a fundamental human motivation. Psychol Bull. (1995)
in Hubei, China. Med Sci Monit. (2020) 26:e924171. doi: 10.12659/MSM. 117:497–529. doi: 10.1037/0033-2909.117.3.497
924171 90. Deci EL, Ryan RM. The “what” and “why” of goal pursuits: human
73. Shaw KA, Chilcott A, Hansen E, Winzenberg T. The GP’s response needs and the self-determination of behavior. Psychol Inq. (2000) 11:227–
to pandemic influenza: a qualitative study. Fam Pract. (2006) 23:267– 68. doi: 10.1207/S15327965PLI1104_01
72. doi: 10.1093/fampra/cml014 91. Van den Broeck A, Ferris DL, Chang C-H, Rosen CC. A review of self-
74. Fagermoen MS. The meaning of nurses’ work: a descriptive study of values determination theory’s basic psychological needs at work. J Manag Stud.
fundamental to professional identity in nursing.: (1996). Dissertation Abstracts (2016) 42:1195–229. doi: 10.1177/0149206316632058
International Section B: The Sciences and Engineering, University of Rhode 92. Shoji S, Dantas de Oliveira Souza NV, Nascimento Pereira Farias S, Cardoso
Island, Kingston RI (1996) 56:4814. Vieira ML, Progianti JM. Proposta de melhoria das condições de trabalho
75. Koh D, Lim MK, Chia SE, Ko SM, Qian F, Ng V, et al. Risk perception and em uma unidade ambulatorial: perspectiva da enfermagem. Escola Anna Nery
impact of severe acute respiratory syndrome (SARS) on work and personal Revista de Enfermagem. (2021) 20:303–9. doi: 10.5935/1414-8145.20160041
lives of healthcare workers in Singapore: what can we learn? Med Care. (2005) 93. Jasper S, Stephan M, Al-Khalaf H, Rennekampff HO, Vogt PM,
43:676–82. doi: 10.1097/01.mlr.0000167181.36730.cc Mirastschijski U. Too little appreciation for great expenditure?
76. Fujiwara T. The role of altruistic behavior in generalized anxiety disorder and Workload and resources in ICUs. Int Arch Occup Environ Health. (2012)
major depression among adults in the United States. J Affect Disord. (2007) 85:753–61. doi: 10.1007/s00420-011-0721-9
101:219–25. doi: 10.1016/j.jad.2006.11.024
77. Fernandes O Jr, Portugal LCL, Alves RCS, Arruda-Sanchez T, Rao Conflict of Interest: The authors declare that the research was conducted in the
A, Volchan E, et al. Decoding negative affect personality trait from absence of any commercial or financial relationships that could be construed as a
patterns of brain activation to threat stimuli. Neuroimage. (2017) 145:337– potential conflict of interest.
45. doi: 10.1016/j.neuroimage.2015.12.050
78. Fernandes O Jr, Portugal LCL, Alves RCS, Arruda-Sanchez T, Publisher’s Note: All claims expressed in this article are solely those of the authors
Volchan E, Pereira MG, et al. How do you perceive threat? It’s and do not necessarily represent those of their affiliated organizations, or those of
all in your pattern of brain activity. Brain Imaging Behav. (2020)
the publisher, the editors and the reviewers. Any product that may be evaluated in
14:2251–66. doi: 10.1007/s11682-019-00177-6
this article, or claim that may be made by its manufacturer, is not guaranteed or
79. Hatton CM, Paton LW, McMillan D, Cussens J, Gilbody S, Tiffin PA.
Predicting persistent depressive symptoms in older adults: a machine learning endorsed by the publisher.
approach to personalised mental healthcare. J Affect Disord. (2019) 246:857–
60. doi: 10.1016/j.jad.2018.12.095 Copyright © 2022 Portugal, Gama, Gonçalves, Mendlowicz, Erthal, Mocaiber, Tsirlis,
80. Baqui P, Bica I, Marra V, Ercole A, van der Schaar M. Ethnic and Volchan, David, Pereira and Oliveira. This is an open-access article distributed
regional variations in hospital mortality from COVID-19 in Brazil: a under the terms of the Creative Commons Attribution License (CC BY). The
cross-sectional observational study. Lancet Glob Health. (2020) 8:e1018- use, distribution or reproduction in other forums is permitted, provided the
26. doi: 10.1016/S2214-109X(20)30285-0 original author(s) and the copyright owner(s) are credited and that the original
81. Ribeiro H, Lima VM, Waldman EA. In the COVID-19 pandemic in publication in this journal is cited, in accordance with accepted academic practice.
Brazil, do brown lives matter? Lancet Glob Health. (2020) 8:e976- No use, distribution or reproduction is permitted which does not comply with these
7. doi: 10.1016/S2214-109X(20)30314-4 terms.

Frontiers in Psychiatry | www.frontiersin.org 14 January 2022 | Volume 12 | Article 752870