Why does this paper matter?
The pandemic highlighted a need for improved virtual access to interventions that can reduce symptoms of depression in older adults. It is unclear which interventions are efficacious for reducing these symptoms. We identified virtual interventions that reduced symptoms of depression in older adults, including cognitive behavioral therapies.
Key points
-
1. Virtual (telephone or videoconference) self-directed or supported cognitive behavioral therapy had statistically significant effects on reducing symptoms of depression in individual studies, but the clinical significance of these findings is unclear.
-
2. Older adults need access to efficacious psychotherapy interventions (e.g. cognitive behavioral therapy, chronic disease support) to reduce symptoms of depression; however, more studies are needed urgently to establish their efficacy via virtual modalities.
Impact statement
This is the first systematic review to identify virtual interventions that reduce symptoms of depression in older community-dwelling adults. We highlight the efficacy of virtual psychotherapy interventions for depression management.
Introduction
Symptoms of depression are one of the most common mental health issues in older adults. Fifteen to 20% of community-dwelling older adults experience ‘substantial symptoms of depression’ (MacCourt and Tourigny-Rivard, Reference MacCourt and Tourigny-Rivard2011). Depressed older adults have poorer quality of life (Doraiswamy et al., Reference Doraiswamy, Khan and Donahue2002), difficulty with day-to-day tasks (Schillerstrom et al., Reference Schillerstrom, Royall and Palmer2008), and higher mortality (MacCourt and Tourigny-Rivard, Reference MacCourt and Tourigny-Rivard2011) compared to those without depression. Efficacious psychotherapies exist to reduce symptoms of depression for older adults with depression (MacQueen et al., Reference MacQueen, Frey and Ismail2016), but a lack of access to these interventions during the pandemic has precluded their use (Raue et al., Reference Raue, McGovern and Kiosses2017; Yang et al., Reference Yang, Li and Zhang2020). Given minimal evidence of efficacy and the risk of adverse events with pharmacologic therapy, psychotherapy is a key initial component of treatment (MacQueen et al., Reference MacQueen, Frey and Ismail2016; Sobieraj et al., Reference Sobieraj, Martinez and Hernandez2019).
Public health measures to minimize COVID-19 spread increased social isolation for older adults, which can worsen mood (Armitage and Nellums, Reference Armitage and Nellums2020; Steinman et al., Reference Steinman, Perry and Perissinotto2020). However, virtual options may facilitate use of interventions in settings where in person access is not feasible. There is increasing support for the use of virtual care in older adults, and rapid uptake during the COVID-19 pandemic has provided an understanding of usefulness (Watt et al., Reference Watt, Fahim and Straus2021; Liu et al., Reference Liu, Goodarzi and Jones2021). A large retrospective cohort from the United States involving 313,516 telehealth visits in adults over 60 years from 2015 to 2019 (Bernstein et al., Reference Bernstein, Ko and Israni2021) found that 84–87% of urgent and nonurgent primary care issues were resolved (Bernstein et al., Reference Bernstein, Ko and Israni2021). For those patients who needed further visits, 95% of these were resolved in less than three visits (either in person or virtual) (Bernstein et al., Reference Bernstein, Ko and Israni2021). The study also noted that certain conditions could be similarly addressed with either phone or videoconference visits (e.g. UTI) (Bernstein et al., Reference Bernstein, Ko and Israni2021).
However, there are concerns around the barriers faced by older adults when trying to access virtual care. A cross-sectional study of 330 patients found that older adults who are frail or do not have a caregiver were less able to access videoconference-based virtual care (Liu et al., Reference Liu, Goodarzi and Jones2021). When examining the accuracy of virtual cognitive testing in a systematic review, many barriers and facilitators to virtual care were identified, including sensory impairment, severe cognitive impairment, physical issues, and issues with technology (Watt et al., Reference Watt, Lane and Veroniki2021).
To inform clinical practice during the COVID-19 pandemic, we conducted a systematic review of randomized controlled trials (RCTs) reporting the efficacy of virtual interventions for reducing symptoms of depression in older adults.
Methods
We registered our systematic review with PROSPERO (CRD42020188465) and disseminated our protocol on Open Science Framework (https://osf.io/6tjcy/). We followed the Cochrane Handbook for Systematic Reviews of Interventions (Higgins and Green, Reference Higgins and Green2011) and reported as per Preferred Reporting Items for Systematic Reviews and Meta-Analysis criteria (Moher et al., Reference Moher, Liberati and Tetzlaff2010; Page et al., Reference Page, McKenzie and Bossuyt2021). All methods were reviewed by stakeholders, including persons with lived experience, and feedback was incorporated.
Search strategy
Our search strategy was developed with an experienced librarian. A second librarian completed a Peer Review of Electronic Search Strategies (PRESS) (McGowan et al., Reference McGowan, Sampson and Salzwedel2016) of the search strategy for all databases. We searched MEDLINE, EMBASE, Cochrane Database of Systematic Reviews, Cochrane Central Register of Controlled Trials, and PsycINFO. We searched references of included studies, systematic reviews, and 76 gray (i.e. difficult to locate and unpublished) literature sources based on the Canadian Agency for Drugs and Technology in Health Grey Matters document and expert opinion (Supplementary File 2). All searches were performed in all languages from inception until July 2021 (Supplementary File 1).
Study selection
We included RCTs comparing the efficacy of any virtual (e.g. telephone, videoconference, or internet-based) nonpharmacologic intervention to usual care or any other virtual nonpharmacologic intervention for reducing symptoms of depression in community-dwelling older adults (defined as ≥60 years old and average age ≥65 years old) with depressive symptoms or disorders at baseline or where depression was measured as an outcome (APA, 2013). We excluded interventions that incorporated both virtual and in-person components, as the focus was on interventions that were entirely administered virtually. We also excluded RCTs where the entire study population had a single specific medical comorbidity (e.g. heart failure), as these populations were too specific. We included studies where the population had multiple comorbidities, as this is representative of older populations. Two reviewers (ZG, JW) independently completed two study screening levels, after a calibration exercise to ensure agreement, conflicts were reviewed by a third party (i.e. title/abstract and full-text papers).
Data extraction and risk of bias assessment
After piloting, ZG and PW independently extracted data from included full-text papers and completed a risk of bias appraisal using the Cochrane Risk of Bias Tool for RCTs (Higgins and Green, Reference Higgins and Green2011, Higgins et al., Reference Higgins, Altman and Gotzsche2011). We extracted the following data: participant characteristics (e.g. age of study population, proportion of female study participants, proportion of participants with symptoms of depression or disorders at RCT baseline, presence of psychiatric comorbidities), study characteristics (e.g. year of publication, authorship, study setting [i.e. urban vs. rural], sample size, study duration, number of RCT intervention arms, inclusion criteria, exclusion criteria), details of intervention, how it was implemented and outcome data (e.g. mean depression scores pre and post intervention) from each intervention group. We extracted outcomes from all follow-up intervals. Where RCTs reported ≥2 scales for the same outcome, we extracted data from all reported scales. We reported outcomes from the commonly used Patient Health Questionnaire-9 (PHQ-9) (Spitzer et al., Reference Spitzer, Kroenke and Williams1999), where available, in order to facilitate comparison between studies. The minimally clinical important difference (MCID) for community-dwelling older adults is 5 points for the PHQ-9 (on a scale from 0 to 27) (Lowe et al., Reference Lowe, Unutzer and Callahan2004). If there were multiple tools used to measure outcomes, we reported tools specific to symptoms of depression or disorders over a subscale. We did not conduct a meta-analysis due to heterogeneity in study populations across studies, comparison groups (e.g. active vs. usual care), and virtual intervention delivery modalities.
Results
Search results
The literature search identified 2633 citations after duplicates were removed (Figure 1). Fifteen RCTs (n = 3100) were included in the systematic review, and one study was reported in two publications (Brenes et al., Reference Brenes, Danhauer and Lyles2017; Brenes et al., Reference Brenes, Danhauer and Lyles2015).
Figure 1. Flow chart.
Study and patient characteristics
All RCTs were published since 2006 (Table 1). The mean age for all participants in included RCTs was 65.1–79.2 years. In included RCTs, the proportion of female participants ranged from 47% to 100%. Included studies were completed predominately in the United States (number [n] = 8) and Australia (n = 4). All RCTs excluded persons with suicidality or severe symptoms of depression or disorders (Supplementary Table 3). Nine studies had participants with co-diagnoses of anxiety and depression.
Table 1. Study and patient characteristics
Abbreviations: Patient Health Questionnaire (PHQ-9, PHQ-8), Montgomery Åsberg Depression Rating Scale (MADRS), or Geriatric Depression Scale (GDS), Centre for Epidemiology Studies Depression Scale (CES-D), Beck Depression Inventory (BDI-I), Cognitive Behavioural therapy (CBT), Hospital Anxiety and Depression Scale-Depression (HADS-D).
a Excluded persons with severe depression based on PHQ-9 >19 or >2 on Suicide item. However, depression nor depressive symptoms were inclusion criteria.
b Included persons with falls, home care, nursing facility living, emergency department visit, hospital admission or depression.
Included RCTs focused on two main groups: persons with symptoms of depression or disorders at baseline (n = 5) or those where symptoms of depression were measured as an outcome only (n = 10).
Risk of bias assessment
Allocation concealment (n = 10 unclear) and blinding procedures (n = 11 unclear; n = 4 high risk) were the most important sources of bias (Supplementary Table 4). Blinding of outcome assessment was variable, with only four studies being at low risk of bias. Most studies had low risk of bias for random sequence generation (n = 13), incomplete outcome reporting (n = 8), and selective reporting (n = 12).
Study outcomes
RCTs where depressive symptoms or disorders were present at baseline (primary outcome).
In five RCTs, where participants had symptoms of depression or disorders at baseline (Table 2), the following therapies were compared to another intervention or usual care: telehealth management of chronic illness (Gellis et al., Reference Gellis, Kenaley and Have2014), problem-solving therapy (Gellis et al., Reference Gellis, Kenaley and Have2014), medication review (Mavandadi et al., Reference Mavandadi, Benson and DiFilippo2015), mental health algorithms (Mavandadi et al., Reference Mavandadi, Benson and DiFilippo2015), and cognitive behavioral therapy (CBT) (Silfvernagel et al., Reference Silfvernagel, Westlinder and Andersson2018; Titov et al., Reference Titov, Fogliati and Staples2016; Titov et al., Reference Titov, Dear and Ali2015). All interventions were delivered over the telephone or internet. The level of health care provider involvement in interventions varied from entirely self-directed to being closely followed by healthcare providers. Health care providers involved were social workers, counsellors, nurses, or psychologists. The cognitive behavioral therapies used included mindfulness, problem- solving skills, behavioral activation, interpersonal skills, and psychoeducation. Self-guided CBT was done at the pace of the individual and sessions with clinicians ranged from 10 to 35 minutes.
Table 2. Change in depression symptom score associated with virtual interventions
Abbreviations: Patient Health Questionnaire (PHQ-9), Montgomery Åsberg Depression Rating Scale (MADRS), or Geriatric Depression Scale (GDS), Centre for Epidemiology Studies Depression Scale (CES-D), Beck Depression Inventory (BDI-I), Cognitive Behavioural therapy (CBT).
a Results of RCT reported across two papers [Brenes, Danhauer & Lyles, Reference Brenes, Danhauer and Lyles2015, Gellis, Kenaley & Have, Reference Gellis, Kenaley and Have2014].
b Reported as Mean Change.
c Reported as significant for the group by time interaction only.
d Reported as significant for the group interaction and the group by time interaction.
e Non-significant change.
f Study reports Hospital Anxiety and Depression Scale – Depression subscale, symptoms were significantly reduced when examining this tool.
* p-value < 0.05, compared to control.
** p-value at 0.05.
The PHQ-9 was the most common tool used in four RCTs (Table 2 and Supplementary Table 5). All RCTs reported immediate post intervention outcomes at 8–12 weeks, and 4 RCTs reported outcomes at 24- or 52-weeks post intervention. Between study clinical and methodological differences precluded meta-analysis.
Of the four RCTs, two showed a statistically significant difference between groups at the initial follow-up (Titov et al., Reference Titov, Dear and Ali2015; Gellis et al., Reference Gellis, Kenaley and McGinty2012; Szczepanska-Gieracha et al., Reference Szczepanska-Gieracha, Cieslik and Serweta2021). Internet CBT with online guidance compared to waitlist control demonstrated efficacy (Titov et al., Reference Titov, Dear and Ali2015). In this study, authors explored an 8 week “Managing Your Mood Course”, focused on five online lessons, with reminders, homework, case stories, and secure emailing with a clinical psychologist (Titov et al., Reference Titov, Dear and Ali2015). This course involved didactic lessons including case learning with skills training (Titov et al., Reference Titov, Dear and Ali2015). When comparing telehealth for chronic illness/depression to nursing check in and education, there was a statistically significant improvement at 3 months, but not at longer follow-up (Gellis et al., Reference Gellis, Kenaley and Have2014). Neither met the between-groups MCID and only had small effect sizes.
Three studies demonstrated a statistically significant group by time interaction between interventions (Mavandadi et al., Reference Mavandadi, Benson and DiFilippo2015; Silfvernagel et al., Reference Silfvernagel, Westlinder and Andersson2018; Shapira et al., Reference Shapira, Yeshua-Katz and Cohn-Schwartz2021). One study examined mental health case management compared to monitoring alone (Mavandadi et al., Reference Mavandadi, Benson and DiFilippo2015). Another examined internet CBT with online guidance by a clinician compared to weekly email support by clinicians (Silfvernagel et al., Reference Silfvernagel, Westlinder and Andersson2018).
Another showed significant change within groups but not between the three comparison groups of: i) internet CBT with orientation and clinician guidance, ii) internet CBT with orientation, and iii) internet CBT alone (Titov et al., Reference Titov, Fogliati and Staples2016). One study examined a Zoom group intervention during COVID-19 (Shapira et al., Reference Shapira, Yeshua-Katz and Cohn-Schwartz2021). In this intervention, participants did twice weekly online guided group sessions with social workers (Shapira et al., Reference Shapira, Yeshua-Katz and Cohn-Schwartz2021). The goal was to provide social interaction and education on cognitive behavioral techniques (Shapira et al., Reference Shapira, Yeshua-Katz and Cohn-Schwartz2021). There was a statistically different change between pre and post loneliness scores (Shapira et al., Reference Shapira, Yeshua-Katz and Cohn-Schwartz2021). None of the studies reporting the PHQ-9 met the MCID.
RCTs where participants do not have baseline symptoms or diagnosis of depression, but where symptoms of depression were measured as an outcome
Ten RCTs measured symptoms of depression solely as an outcome (Table 2 and Supplementary Table 5). Interventions focused on CBT investigated a range of support from an initial orientation to weekly clinician support and included various aspects of therapy such as relaxation (Supplementary Table 5 for details). Most studies (n = 5) looked at populations with anxiety (Brenes et al., Reference Brenes, Danhauer and Lyles2017; Brenes et al., Reference Brenes, Danhauer and Lyles2015; Dear et al., Reference Dear, Zou and Titov2013; Gould et al., Reference Gould, Kok and Ma2019; Jones et al., Reference Jones, Hadjistavropoulos and Soucy2016; Brenes et al., Reference Brenes, Miller and Williamson2012). The following therapies were studied: telephone and internet delivery of CBT (Brenes et al., Reference Brenes, Danhauer and Lyles2017; Brenes et al., Reference Brenes, Danhauer and Lyles2015; Jones et al., Reference Jones, Hadjistavropoulos and Soucy2016; Brenes et al., Reference Brenes, Miller and Williamson2012; Dear et al., Reference Dear, Zou and Ali2015; Read et al., Reference Read, Sharpe and Burton2020); nondirective supportive therapy (Brenes et al., Reference Brenes, Danhauer and Lyles2017; Brenes et al., Reference Brenes, Danhauer and Lyles2015); psychologist and breathing exercises (Gould et al., Reference Gould, Kok and Ma2019); interactive website (Gustafson et al., Reference Gustafson, Kornfield and Mares2021); online group intervention (Shapira et al., Reference Shapira, Yeshua-Katz and Cohn-Schwartz2021); support groups (Hartke and King, Reference Hartke and King2003); and assessment and follow-up of physical symptoms/quality of life/social support (Kornblith et al., Reference Kornblith, Dowell and Herndon2006). Five studies reported outcomes with the PHQ-9, one with PHQ-8, one with Centre for Epidemiology Studies Depression Scale, two with the Beck depression inventory (BDI-I), and one with the Geriatric Depression Scale (GDS). Outcomes were reported between 4 and 60 weeks and three reporting at 24 months (Brenes et al., Reference Brenes, Miller and Williamson2012; Read et al., Reference Read, Sharpe and Burton2020; Hartke and King, Reference Hartke and King2003). All studies showed improvement in depression scores and none reported harms. When examining internet CBT with clinician guidance compared to waitlist control, there was a statistically significant outcome at 8 weeks (Dear et al., Reference Dear, Zou and Ali2015). Another study demonstrated a significant group and group by time interaction for guided internet CBT compared to usual care (Read et al., Reference Read, Sharpe and Burton2020).
Discussion
The need for virtual interventions that can reduce symptoms of depression and depressive disorders in older adults transcend the COVID-19 pandemic. Challenges with transportation, mobility, living rurally, lack of access to services locally, and cost all lead to poor access to psychotherapy for community-dwelling older adults (Wuthrich and Frei, Reference Wuthrich and Frei2015). Using virtual interventions helps to bridge some of the gaps to access for older adults (Gould and Hantke, Reference Gould and Hantke2020).
We identified 15 RCTs evaluating the efficacy of virtual interventions for reducing symptoms of depression in older adults. These studies demonstrate that virtual interventions are feasible, and in some studies demonstrated efficacy. Harm was not reported in any study. However, there was insufficient evidence to establish whether these interventions lead to clinically meaningful outcomes or durable results for patients. Virtual interventions were delivered via telephone or internet and had varied degrees of involvement from clinicians.
CBT is a well established and effective intervention for depression in person (Olthuis et al., Reference Olthuis, Watt and Bailey2016; Gotzsche and Gotzsche, Reference Gotzsche and Gotzsche2017; Wuthrich and Rapee, Reference Wuthrich and Rapee2013; Jayasekara et al., Reference Jayasekara, Procter and Harrison2015; Huibers, Reference Huibers2011; Hall et al., Reference Hall, Kellett and Berrios2016) and appears to adapt and be feasible in the virtual setting. However, the virtual setting does not completely eliminate issues of access to CBT associated with cost and the need for clinician involvement (Wuthrich and Frei, Reference Wuthrich and Frei2015).
There are factors to consider regarding feasibility of different virtual care modalities. A qualitative interview study looked at virtual care in a urban geriatric medicine clinic (not specific to interventions for depression) (Watt et al., Reference Watt, Fahim and Straus2022). Eight major barriers were identified in this study including the challenges associated with rapid uptake of virtual care due to COVID-19, providing virtual care to a medically complex population, the importance of ensuring caregivers are involved, difficulty with understanding the accuracy of virtual assessments compared to in person, issues with inequitable access, difficulties with changing clinic processes, technology issues, and the importance of technology uptake by older adults (Watt et al., Reference Watt, Fahim and Straus2022). These barriers, although focused on geriatric medicine clinic and not depression interventions, are likely transferable to other virtual interventions. Especially given the overlap with the barriers identified in a systematic review of the accuracy virtual cognitive tools (Watt et al., Reference Watt, Lane and Veroniki2021).
There are additional social and health factors that impact accessibility of virtual care. Older adults without an available caregiver, or who are more frail, are less able to access videoconferencing options (Liu et al., Reference Liu, Goodarzi and Jones2021). Video-conferencing provides more information to the health care provider and more direct contact for patients, and while we don't have evidence to suggest that this is necessarily better than telephone-based virtual care; it does have advantages (Liu et al., Reference Liu, Goodarzi and Jones2021). Virtual care does enable a larger reach for programming, but there is a need to consider issues like individual access or capability with technology for any client (Watt et al., Reference Watt, Fahim and Straus2022). But providers must further consider the issues facing rural persons specifically such as internet bandwidth (Watt et al., Reference Watt, Fahim and Straus2022). Along with these there is very little literature speaking to the potential applicability of virtual care across different languages or ethnicities and how this may impact patient outcomes (Liu et al., Reference Liu, Goodarzi and Jones2021; Watt et al., Reference Watt, Lane and Veroniki2021; Reference Watt, Fahim and StrausWatt et al., 2022). These issues can, however, be further examined and services optimized to ensure intersectional factors are considered when delivering virtual care.
Our study has several strengths. We completed a rigorous search of indexed databases and grey literature. Despite this, we were unable to complete a meta-analysis due to important clinical heterogeneity. There were several limitations to included studies such as unclear or high risk of bias from participant and assessor blinding. All trials excluded persons with severe depression and persons experiencing suicidality; thus, it is unclear if these interventions are suitable or effective in these patient groups. No studies included persons with cognitive impairment or substance use disorders. Studies did not report diversity in race and ethnicity hence we cannot extrapolate to these populations.
Conclusion
Even prior to the COVID-19 pandemic, there was a need for improved accessibility to psychotherapy and interventions such as CBT for depression management in older adults. However, there is a need for further studies examining the efficacy and cost of virtual care for treating symptoms of depression and disorders in older adults. Future work should examine other barriers to virtual care, ensure inclusive participant recruitment, and establish efficacy for virtual psychotherapy for depression. A deeper understanding regarding the patient experience and perceptions of virtual care for mental illness is needed.
Conflict of interest
ZI reports conflicts are that they have received honoraria from Lundbeck and Otsuka, outside of the submitted work.
Source of funding
ZG is funded by the Canadian Institutes of Health Research, Hotchkiss Brain Institute, and O’Brien Institute of Public Health. SES is funded by a Tier 1 Canada Research Chair in Knowledge Translation. ACT is funded by a Tier 2 Canada Research Chair in Knowledge Synthesis.
Description of authors’ roles
All authors were involved in the development of the protocol, research questions, revisions, and final manuscript.
ZG, PW, JW, and JHL were involved in abstract, full text, data extraction, and risk of bias assessment.
Disclosures
This study was funded by the Canadian Institutes of Health Research (CIHR): https://cihr-irsc.gc.ca/e/52108.html. This was presented at a webinar with the BrainXChange in September 2020: https://brainxchange.ca/Public/Events/Archived-Webinars-Events/2020/How-can-we-best-help-older-adults-with-depression. This work was posted as a report on the CIHR website: https://cihr-irsc.gc.ca/e/52057.html.
Acknowledgments
This study was funded by the Canadian Institute for Health Research. We would like to thank Diane Lorenzetti and Jessie McGowan who are expert information scientists who aided in the development of the search.
We are grateful for the guidance of Dr Barbara Liu University of Toronto and the Regional Geriatric Program of Ontario as well as Charlene Retzlaff, social worker in the Family Caregiver Centre at Alberta Health Services.
Supplementary material
To view supplementary material for this article, please visit https://doi.org/10.1017/S1041610222000412
Open access
