Identifying the research question

A single research question was identified as the primary focus of the review: How has the COVID-19 pandemic impacted the mental health symptoms of individuals with pre-existing depressive, anxiety, or specified stressor-related (i.e., posttraumatic stress, acute stress) disorders/issues?

Identifying relevant studies

Articles were identified through searches using database-specific subject headings and keywords in natural language in the following databases: Medline (including Epub ahead of print, in-process and other non-indexed citations), Embase, APA PsycInfo, and Cumulative Index to Nursing & Allied Health Literature (CINAHL).

A medical librarian (TR) developed the search strategies with input from the review team and conducted all searches on October 29, 2020 and again on March 3, 2021. Mental health conditions considered were: pre-existing depressive, anxiety, or specified stressor-related (i.e., posttraumatic stress, acute stress) disorders/issues. To capture articles on pre-existing mental health conditions that may not be described as such, we used natural language search terms such as “improve”, “lessen”, “worsen”, and “exacerbate”, which imply a change over time from a pre-existing state. Due to challenges involved in employing the same terms (i.e., depression and anxiety) in two different concepts within the same search strategy (see inclusion criteria 1 and 3 below), we also designed the strategy to query pre-existing depression, anxiety or stressor-related disorders separately from all other pre-existing mental health conditions. Publication year limits applied were 2019-present, reflecting the onset of the COVID-19 pandemic. No language limits were applied. The full Medline search strategy can be found in Table 1. Records returned from this search were managed in Covidence systematic review software.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 1. Ovid medline database search strategy: Epub ahead of print, in-process & other non-indexed citations, ovid MEDLINE® daily and ovid MEDLINE® <1946-present>.

https://doi.org/10.1371/journal.pone.0295496.t001

Selecting eligible studies

The review of potentially relevant studies proceeded in two stages: (1) title and abstract review, and; (2) full paper review. To reduce potential bias, the title and abstract of each record were screened by two independent reviewers to identify articles potentially relevant to the scoping review. Conflicts between independent reviewers regarding eligibility of a study for inclusion were discussed and resolved by a minimum of three reviewers, including the Principal Investigator. In determining relevance of articles focused on specific diagnoses, the Diagnostic and Statistical Manual of Mental Disorders (DSM-5) [9] was used; all diagnoses listed under anxiety disorders and depressive disorders were included, as were posttraumatic stress disorder (PTSD) and acute stress disorder.

Based on screening of titles and abstracts, potentially relevant articles were read in full by two independent reviewers. To be included in the final analysis, data related to the mental health symptoms or diagnoses of interest could not be intermixed or merged with data related to other mental health issues that were not relevant to our research question. For example, if the effects of the pandemic were assessed in patients with anxiety, depression or schizophrenia, the article could only be included in the review if analyses examined patients with anxiety and/or depression separately from patients with schizophrenia. A few exceptions were made for studies where only a small minority of the sample was diagnosed with a condition outside of the inclusion criteria; these cases are identified in the data extraction table (see Table 1). Since suicidal ideation is a possible symptom of depression, but is not exclusive to this diagnosis, articles where suicide-related symptoms (e.g., ideation, attempts) were included as a relevant outcome variable were included in the review, while articles in which suicide-related symptoms were listed as the sole pre-existing mental health condition were excluded. In the full text review stage, conflicts between two independent reviewers regarding eligibility of a study for inclusion were resolved by the Principal Investigator. To be included in the final analysis, an article had to be selected in the full paper review stage by at least two reviewers, one of whom had to be the Principal Investigator.

Charting the data and synthesizing the results

Study details (i.e., location, design and methodology, sample or population, outcome measures, and key findings) were extracted from included studies by one reviewer and confirmed by the Principal Investigator. Data were collated using narrative synthesis.

Geographical representation

Selected articles were drawn from 26 countries, across five continents (see Table 3). Europe contributed the greatest number of articles (n = 29), with most coming from Italy (n = 6) and Spain (n = 5). This was followed by Asia (n = 16), with most articles coming from China (n = 9), and by North America (n = 15), with most articles coming from the United States (n = 13). These countries were among those with the highest COVID-19 case counts in the first few months of the pandemic [79]. In addition, one article included data from Turkey, one included data from Cyprus, and another included data from Russia, which are countries on the border of Europe and Asia. These numbers have been excluded from the continental counts. Only two studies reported inclusion of data from more than one country. Specifically, Asmundson et al. [19] included data from both Canada and the USA, and Thombs et al. [73] included data from the USA, the UK, Canada, and France.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 3. Geographical representation of included studies (n = 66).

https://doi.org/10.1371/journal.pone.0295496.t003

Research design

The majority of studies adopted a cross-sectional design (i.e., a single data collection period following onset of COVID-19 cases; n = 42), perhaps because this study design would allow for the most immediate findings. The next most frequently reported study design was a two time-point pre-post design (n = 13). In some cases, these studies were conducted by teams who had relevant pre-pandemic data available (e.g., [18]); in others, participants were asked to report on pre-pandemic mental health retrospectively (e.g., [66]). Other study designs included longitudinal (i.e., more than two data points; n = 1), matched case-control (i.e., two existing groups are compared on the basis of a potential causal attribute; n = 4), chart review (i.e., a review of pre-recorded, patient-centred data; n = 3), and case study (e.g., intensive study of a single patient; n = 3). See Table 4 for a summary of study designs used by relevant studies. Several studies planned to include subsequent data collection phases (e.g., [54,69]). Among the cross-sectional and matched case-control studies (n = 46), the vast majority made some form of comparison between individuals with pre-existing depressive, anxiety, or specified stressor-related disorders or issues and: (a) individuals with no history of mental illness or those who scored below a threshold value on a screening instrument (n = 35) or; (b) those with other mental health diagnoses (n = 10). Among the pre-post and longitudinal studies (n = 14), four examined risk of current psychiatric symptoms among those with versus without history of mental illness, but most studies reported on changes in severity of symptoms (n = 4) or whether the severity of symptoms at time 1 predicted the severity of symptoms at time 2 (n = 6).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 4. Research design, methodology, sample source, pre-existing disorders/symptoms, and measured outcomes of included studies (n = 66).

https://doi.org/10.1371/journal.pone.0295496.t004

Methodology

Most studies surveyed research participants, primarily via online survey (n = 50), although several employed alternative approaches; these included data collected through paper-and-pencil surveys (n = 2), over telephone or videoconference (n = 9), and in-person interviews (n = 3). Among cross-sectional surveys, a few studies (n = 3) allowed participants to choose how they accessed the survey during the pandemic (e.g., online, paper-and-pencil, telephone, in-person interview; [30,48,59]). One cross-sectional study used “a convergent mixed-methods approach” ([67], p. 4) including both in-depth qualitative telephone interviewing and online survey. Hölzle et al. [43] conducted a cross-sectional survey but did not specify the method by which data were collected. See Table 4 for a summary of methodologies used by relevant studies.

Sample/Population

Participants were drawn from a diversity of sources. Twenty-six studies were conducted with samples (or control/comparison samples) clearly drawn from the general population. Of these, five studies used probabilistic samples or quota-based sampling (e.g., [19,23]), four of which recruited participants from existing panels or ongoing studies and one recruited through random digit dialling. Of the studies drawn from the general population that were non-probabilistic (e.g., [56,61]), most used social media, online or conventional media advertising or snowballing techniques to recruit participants; however, some recruited all or a portion of their participants from existing panels or ongoing studies. Eighteen studies were conducted with samples of mental health inpatients or outpatients recruited from hospitals, clinics, mental health programs or ongoing studies. Other special samples, either as a primary or comparison group, included university students (n = 2; [35,65]), healthcare workers (n = 8; e.g., [77,78]), patients currently or previously diagnosed with COVID-19 (n = 2; [30,40]), and patients with other medical conditions such as breast cancer, Parkinson’s disease, and current or recent pregnancy (n = 10; e.g., [7,44,53]). These special samples were typically recruited from existing institutions (e.g., hospitals, universities), although advertising and snowballing techniques were used to recruit half of the studies of patients with other medical conditions. Table 4 summarizes the various sources from which relevant studies drew their study samples.

Pre-existing disorders/symptoms

Among the included studies, the most commonly reported pre-existing category of disorders or symptoms was depressive disorders (n = 51; e.g., major depressive disorder), followed closely by anxiety disorders (e.g., n = 40; generalized anxiety disorder, panic disorder). Seven studies reported a pre-existing condition that could be labelled as depression or anxiety (e.g., affective disorder). A total of 62 studies included participants with a history of or pre-pandemic experience of depression and/or anxiety or related symptoms. Far fewer studies examined the stressor-related disorders of PTSD or acute stress disorder or related symptoms (n = 13), and most of these studies (n = 9) also included participants with a history of or pre-existing depressive and/or anxiety symptoms. See Table 4 for a summary of pre-existing disorders or symptoms examined by relevant studies. Most studies (n = 48) focused on individuals with a diagnosed disorder, although in about half of these studies the diagnoses were reported by the participant themselves (n = 21). For this reason, diagnoses provided in studies of patient populations, chart reviews, or case studies were likely more reliable. Other studies (n = 19) relied on self-reporting of symptoms or screening measures to identify individuals with a history of mental health disorder or pre-existing symptoms. Aragona et al. [18] included symptom measures in a sample of diagnosed patients.

Outcome measures

Most studies (n = 46) reported more than one measured outcome. Across these studies, 21 different combinations of outcomes were found, with no more than four studies sharing the same combination of outcomes. The only exception was the most frequently reported combination of depression and anxiety. Two studies included an outcome measure of depression alone and three included a measure of anxiety alone. Forty studies included measures of both depression and anxiety, with 22 of these studies including additional outcomes as well. Depression and anxiety were assessed using a broad diversity of predominantly well-validated and reliable measures (e.g., Brief Symptom Inventory (BSI-18); Depression, Anxiety, and Stress Scale (DASS-21); Generalized Anxiety Disorder (GAD-7) Scale; one of several versions of the Patient Health Questionnaire (PHQ)). In three of these studies, analyses treated anxiety and depression as a single construct [23,42,53]. Either alone or in combination with other outcome measures, about one quarter of studies included measures of general stress or distress (n = 16) and another quarter of studies included measures of PTSD symptoms (n = 14); only four studies included both types of stress-related measures [6,37,40,75]. General stress or distress was most frequently measured by the DASS-21 stress subscale (n = 7) or the Perceived Stress Scale (n = 3), and PTSD symptoms were most frequently measured by the PTSD Checklist (PCL-5 or PCL-C; n = 8) or the Impact of Event Scale (IES; n = 3). Either alone or in combination with other outcomes, about one quarter of studies also measured anxiety specific to the pandemic (n = 14), relying on items developed specifically for their survey or interview or on newly developed screening scales such as the COVID Stress Scale (e.g., [19,48]) or Coronavirus Anxiety Scale (e.g., [56,60]). Eleven studies examined suicidal thoughts and behaviours, commonly drawing on patient files [41,45,46] or responses to the Columbia Suicide Severity Rating Scale [22,54,55]. Five studies developed items asking participants to rate perceived impact of the pandemic on their mental health (e.g., [34,57]) and four studies included a measure of symptom severity (i.e., Clinical Global Impression–impression and severity subscales; e.g., [47,59]). Table 4 includes a summary of measured outcomes reported by relevant studies.

Key findings of eligible studies

In reviewing study findings, it was clear that the vast majority of studies found an association between a pre-existing mental health issue and increased vulnerability to adverse mental health outcomes during the COVID-19 pandemic. Depending on study design, these studies noted either: an increase in symptoms during the pandemic among those with pre-existing symptoms; an increased risk of developing symptoms among those with a history of the condition or pre-existing symptoms, or; a higher prevalence or severity of symptoms among those with a history of the condition or pre-existing symptoms compared to either those without such a background or those with an alternative mental or physical health condition.

Among individuals with a history of or pre-existing depression or depressive symptoms, the greatest number of studies detected an increase in or greater risk of depression (n = 20), followed closely by an increase in or greater risk of anxiety (n = 13). Pre-existing depression or depressive symptoms were also associated with increased or greater risk of symptoms of stress (n = 5), COVID-related stress/anxiety (n = 3), PTSD (n = 2), suicidal thoughts or behaviour (n = 5), and a perception that the pandemic had impacted one’s mental health (n = 1). Results were similar among individuals with a history of or pre-existing anxiety or anxiety symptoms, for whom the greatest number of studies identified increased or greater risk of anxiety (n = 13), followed by an increase in pandemic-specific stress/anxiety (n = 6). Associations with increased symptoms of depression (n = 4), suicidal thoughts or behaviour (n = 3), stress (n = 1), and PTSD (n = 1) were also found. Additional studies of pre-existing depression or depressive symptoms (n = 5) and of pre-existing anxiety or anxiety symptoms (n = 1) identified a positive association with during-pandemic symptoms of depression, anxiety, or stress (treated as a single outcome). Analyses in several studies considered individuals with pre-existing depression or anxiety (treated as a singular condition), or depressive or anxiety symptoms. Such studies identified an association with depressive (n = 9), anxiety (n = 9), and stress (n = 8) symptoms, as well as PTSD symptoms (n = 4), pandemic-related stress/anxiety (n = 4), perceived negative impact on one’s mental health (n = 3), depressive or anxiety symptoms (treated as a single outcome; n = 1), and suicidal thoughts or behaviour (n = 1). Analyses in two additional studies (i.e., [66,68]) treated individuals with pre-existing depression or anxiety or PTSD (treated as a singular condition) and found associations with increased risk of depressive, anxiety, and PTSD symptoms during the pandemic.

A few studies provided a comparison of during-pandemic symptoms among individuals with a history of or pre-existing depression or depressive symptoms and those with a history of or pre-existing anxiety or anxiety symptoms. Asmundson et al. [19] reported that a group with anxiety-related disorders had higher levels of anxiety and COVID-19 related stress/anxiety than a mood-disorder group, but similar levels of depression. Costa et al. [27] found that within an online peer support community of individuals living with mental illness, those with an anxiety disorder reported the most stresses or concerns about the COVID-19 pandemic. Sorokin et al. [71] found that perceiving COVID-19 as a threat to one’s own life was more common among those reporting an anxiety disorder than a mood disorder. However, based on data from ongoing cohort studies, Pan et al. [57] found that perceived mental health impact, fear of COVID-19, and changes in symptoms from before to during the pandemic were largely similar across various mood and anxiety disorders.

Aside from studies focusing on pre-existing depression or anxiety, several studies found that a history of pre-existing trauma or PTSD, or related symptoms, was associated with increased risk of depressive (n = 2), anxiety (n = 1), and PTSD (n = 3) symptoms and suicidal thoughts or behaviour (n = 2) during the pandemic. Also, Saraswathi et al. [65] reported that higher pre-pandemic stress scores were associated with higher levels of stress during the pandemic.

Findings both within and across studies were not entirely consistent; a small number of studies (n = 13) reported no difference in vulnerability for at least one tested symptom among those with pre-existing depression, anxiety, or stressor-related disorders or symptoms. However, all but two of these studies simultaneously found evidence of increased vulnerability to other symptoms (e.g., [7,28,35]). Only five studies found any evidence of a reduced vulnerability for mental health symptoms among individuals with pre-existing mental health issues, or among individuals with more severe versus less severe pre-existing symptoms (i.e., [34,57,64,72,73]). However, as with the few cases of null effects, findings of reduced vulnerability were typically found for select symptoms only and accompanied demonstrated effects of increased vulnerability (e.g., [34,57]). Sun et al. [72] is an exception; among those with a mental health diagnosis, no to mild levels of anxiety or depressive symptoms before the pandemic (measured retrospectively) were associated with increased risk for worsened anxiety and depressive symptoms during the pandemic. Of the studies that identified a reduced vulnerability for mental health symptoms among those with a pre-existing mental health issue or more severe symptoms, one adopted a cross-sectional design, one adopted a matched case-control design, and three were pre-post studies.

Strengths and limitations

These results are generally consistent with other reviews examining this issue [80–82]. However, the current review assessed the first full year of the COVID-19 pandemic as opposed to the first several months, which is a limitation of the early reviews (e.g., [81]). The current review also focused on depressive, anxiety, and specified stressor-related disorders, all examples of ‘common mental disorder’. This is important because different disorders are associated with different symptoms and treatment needs, which may be differentially impacted by the pandemic and associated public health restrictions. There is emerging evidence to suggest that the mental health impact of the pandemic differed for individuals with severe mental disorder (e.g., a psychotic disorder) compared to those with a common mental disorder [37]. Depressive, anxiety, and specified stressor-related disorders were selected as the focus of the current review because literature from previous disease outbreaks and early literature of the general population at the onset of the COVID-19 pandemic identified an increase in symptoms typical of these disorders [11–13].

Limitations of the review included its focus on DSM-5 classifications [9] to define depressive and anxiety disorders. ICD-10 classifications [10], which are commonly used by mental health professionals outside of North America, do not align perfectly with the DSM-5. Thus, the review would likely have included additional articles had the DSM-5 not been used to interpret inclusion criteria; however, it cannot be determined if the results of the review would have differed substantially. Although the current review focused on specific mental health issues, it was not restricted to mental health patients only; it included studies of individuals with a history of or pre-existing mental health disorder or self-reported symptoms associated with these disorders, allowing assessment among individuals without a formal or current diagnosis. There was no requirement that studies adopt diagnostic instruments rather than rely on brief mental health screening scales, which are intentionally sensitive, resulting in higher risk of false-positive ratings, overestimation of prevalence, and lack of concordance with functional impairment [83]. Likewise, where diagnoses were considered, there was no requirement that diagnoses be confirmed by clinicians or chart review rather than be self-reported by patients. Given the limited number of studies that included diagnostic instruments or clinician-confirmed diagnoses, it was important to use broad inclusion criteria in order to map relevant literature that was available in the first year of the pandemic. A related limitation that is characteristic of scoping review methodology is the absence of quality assessment, which would specify the level of methodological rigour associated with each study. While scoping reviews provide a map or an overview of the evidence, they do not aim to produce a synthesized answer to a particular question. For this reason, assessment of methodological limitations or risk of bias in the evidence is not conducted as part of a scoping review [84,85].

Implications

Research findings summarized by the review have important implications. Tracking the worsening of symptoms during the COVID-19 pandemic among individuals with a pre-existing mental health issue can be used to promote action in improving the mental healthcare system, better equipping it to address concerns from its existing patient population and mitigate further mental deterioration [36]. In Canada, access to publicly funded mental healthcare was limited prior to the pandemic, which only exacerbated the situation. Mental health research funding was not proportionate to the burden of mental illness on the population [86]. The rise in depressive, anxiety, and stressor-related symptoms in the population during the pandemic [1,2,13], particularly among those with pre-existing mental health issues as indicated by this review, underscores the need to dedicate sufficient resources to the mental healthcare system for treatment services. Moreover, clinical surveillance of mental health symptoms during the pandemic and beyond is essential to empowering primary care providers, family, caregivers, and patients themselves to target screening and prevention efforts to those most at risk of a mental health crisis. This can only be achieved through appropriate funding for monitoring and analysis of health system data and mental health research more broadly. During the pandemic, many mental health treatments shifted from in-person patient-facing care to virtual care [86]. Through both self-report and review of health service administrative data, future research should also assess the extent to which patients were able to access needed care and evaluate the effectiveness of its virtual delivery.

Summated characteristics of relevant studies conducted in the first year of the pandemic also have important implications. With data from just 26 countries and five continents worldwide, it is clear that research on symptoms of common mental disorder among those with pre-existing mental health issues lacked geographical representation. Moreover, the bulk of studies were conducted in high-income countries, with little exploration in middle-income countries and no exploration in low-income countries. The review also found that studies reported primarily cross-sectional online surveys, often examining general population samples. More diverse study designs from a broader geographic range, including more low- and middle-income countries, are needed.

There has been an unprecedented proliferation of research related to the COVID-19 pandemic, including studies of its impact on mental health. For this reason, further reviews of this expansive literature will be essential to track and synthesize relevant findings and ensure that they appropriately inform policy and practice. A review of research published in subsequent years of the pandemic is warranted, and may uncover a shift in study design, methodology, study instruments, and sample, as researchers became better positioned to conduct more complex research through government funding for pandemic focused study. This anticipated rise in diversity, complexity, and rigour of pandemic-related studies will augment the potential value of assessing study quality in future reviews, supporting adoption of a systematic review approach [84]. Additional reviews should be conducted on research of other mental health diagnoses, to assess if individuals afflicted are also more vulnerable to worsening of related symptoms and/or to depressive, anxiety, and stressor-related symptoms, as assessed here. Finally, a review should be conducted to examine research of the intersection of social determinants of health with pre-existing mental illness as a vulnerability to the mental health impacts of the pandemic. Data have emerged to suggest that marginalized groups, including those who are racialized or economically disadvantaged, have experienced greater mental health effects of the pandemic [87,88]. This was beyond the scope of the current review, but could further refine our ability to target screening, prevention, and treatment within the mental healthcare system.

Conclusion

The COVID-19 pandemic has had a profound impact on the mental health of the global population. Findings of the current review suggest that individuals with pre-existing mental health issues were at greater risk of adverse mental health impacts of the pandemic. These findings underscore the need for improved support of the mental healthcare system and continued mental health research, including reviews of pandemic effects on individuals with other mental health diagnoses and reviews of research published in subsequent years of the pandemic.