Scolaris Content Display Scolaris Content Display

Interventions for treating anxiety after stroke

Collapse all Expand all

Background

Approximately 20% of stroke patients experience clinically significant levels of anxiety at some point after stroke. Physicians can treat these patients with antidepressants or other anxiety‐reducing drugs, or both, or they can provide psychological therapy. This review looks at available evidence for these interventions. This is an update of the review first published in October 2011.

Objectives

The primary objective was to assess the effectiveness of pharmaceutical, psychological, complementary, or alternative therapeutic interventions in treating stroke patients with anxiety disorders or symptoms. The secondary objective was to identify whether any of these interventions for anxiety had an effect on quality of life, disability, depression, social participation, caregiver burden, or risk of death.

Search methods

We searched the trials register of the Cochrane Stroke Group (January 2017). We also searched the Cochrane Central Register of Controlled Trials (CENTRAL; the Cochrane Library; 2017, Issue 1: searched January 2017); MEDLINE (1966 to January 2017) in Ovid; Embase (1980 to January 2017) in Ovid; the Cumulative Index to Nursing and Allied Health Literature (CINAHL; 1937 to January 2017) in EBSCO; and PsycINFO (1800 to January 2017) in Ovid. We conducted backward citation searches of reviews identified through database searches and forward citation searches of included studies. We contacted researchers known to be involved in related trials, and we searched clinical trials registers for ongoing studies.

Selection criteria

We included randomised trials including participants with a diagnosis of both stroke and anxiety for which treatment was intended to reduce anxiety. Two review authors independently screened and selected titles and abstracts for inclusion.

Data collection and analysis

Two review authors independently extracted data and assessed risk of bias. We performed a narrative review. We planned to do a meta‐analysis but were unable to do so as included studies were not sufficiently comparable.

Main results

We included three trials (four interventions) involving 196 participants with stroke and co‐morbid anxiety. One trial (described as a 'pilot study') randomised 21 community‐dwelling stroke survivors to four‐week use of a relaxation CD or to wait list control. This trial assessed anxiety using the Hospital Anxiety and Depression Scale and reported a reduction in anxiety at three months among participants who had used the relaxation CD (mean (standard deviation (SD) 6.9 (± 4.9) and 11.0 (± 3.9)), Cohen's d = 0.926, P value = 0.001; 19 participants analysed).

The second trial randomised 81 participants with co‐morbid anxiety and depression to paroxetine, paroxetine plus psychotherapy, or standard care. Mean levels of anxiety severity scores based on the Hamilton Anxiety Scale (HAM‐A) at follow‐up were 5.4 (SD ± 1.7), 3.8 (SD ± 1.8), and 12.8 (SD ± 1.9), respectively (P value < 0.01).

The third trial randomised 94 stroke patients, also with co‐morbid anxiety and depression, to receive buspirone hydrochloride or standard care. At follow‐up, the mean levels of anxiety based on the HAM‐A were 6.5 (SD ± 3.1) and 12.6 (SD ± 3.4) in the two groups, respectively, which represents a significant difference (P value < 0.01). Half of the participants receiving paroxetine experienced adverse events that included nausea, vomiting, or dizziness; however, only 14% of those receiving buspirone experienced nausea or palpitations. Trial authors provided no information about the duration of symptoms associated with adverse events. The trial of relaxation therapy reported no adverse events.

The quality of the evidence was very low. Each study included a small number of participants, particularly the study of relaxation therapy. Studies of pharmacological agents presented details too limited to allow judgement of selection, performance, and detection bias and lack of placebo treatment in control groups. Although the study of relaxation therapy had allocated participants to treatment using an adequate method of randomisation, study recruitment methods might have introduced bias, and drop‐outs in the intervention group may have influenced results.

Authors' conclusions

Evidence is insufficient to guide the treatment of anxiety after stroke. Further well‐conducted randomised controlled trials (using placebo or attention controls) are required to assess pharmacological agents and psychological therapies.

PICOs

Population
Intervention
Comparison
Outcome

The PICO model is widely used and taught in evidence-based health care as a strategy for formulating questions and search strategies and for characterizing clinical studies or meta-analyses. PICO stands for four different potential components of a clinical question: Patient, Population or Problem; Intervention; Comparison; Outcome.

See more on using PICO in the Cochrane Handbook.

Interventions for treating anxiety after stroke

Review question

To determine whether any treatments might reduce the symptoms of anxiety, and subsequently improve quality of life, for people who have had a stroke.

Background

Anxiety after stroke occurs frequently and can be treated with antidepressants or other anxiety‐reducing drugs, or both, or with psychological therapy.

Study characteristics

Evidence is current to January 2017. We found three studies with 196 stroke survivors who had received a diagnosis of anxiety. One study assessed the effect of a relaxation CD used five times a week for one month for participants with a diagnosis of anxiety. Two studies assessed the use of antidepressants in participants who had both anxiety and depression.

Key results

One study found that participants were less anxious three months after using a relaxation CD when compared with those who were given no therapy. One study reported that participants were less anxious when treated with an antidepressant medicine (paroxetine), or with paroxetine and psychotherapy, than with standard care. This study reported that half of the participants receiving paroxetine experienced side effects that included nausea, vomiting, or dizziness. The third study also reported that participants were less anxious when treated with an antidepressant (buspirone hydrochloride) than with standard care, and only 14% of those receiving buspirone hydrocholoride reported nausea or palpitations.

Quality of the evidence

We judged that the quality of this evidence was very low. Studies were few and each included a small number of participants. Studies assessing antidepressants did not include comparison with a placebo drug, and information in both study reports was insufficient to permit assessment of whether other biases had been introduced. The study of relaxation therapy was very small, with loss of two participants who used the CD, and the study recruitment process may have attracted participants who had a positive bias towards psychological therapies.

Conclusion

Current evidence is insufficient to guide the treatment of anxiety after stroke. Additional well‐conducted randomised trials are needed.

Authors' conclusions

Implications for practice

Currently, evidence is insufficient to guide practice in treatment of anxiety after stroke. The pharmaceutical therapies evaluated indicate that, when compared with standard care, medication may be an effective approach for reducing anxiety symptoms in stroke patients with co‐morbid anxiety and depression. The clinical significance of this decrease is unclear, as study authors did not provide any information about the proportion of study participants no longer meeting the anxiety criteria. Research indicates that a reduction of more than 50% on the Hamilton Anxiety Scale is indicative of tangible improvement in the level of anxiety (Ye 2006). However, the quality of the evidence for pharmaceutical therapies in this review is very low. The relaxation therapy evaluated in this review was examined in a small pilot study, and although study authors reported a statistically significant reduction in anxiety at three months after use of an autogenic relaxation CD for one month, risk of bias inherent in the study design was high, and not all participants used the CD as often as directed.

Implications for research

Given the high prevalence of anxiety after stroke, placebo‐controlled or attention‐control trials are needed to identify effective treatments for this condition, as it can have a negative impact on other aspects of life. Future research evaluating interventions to treat post‐stroke anxiety should assess outcomes such as quality of life and caregiver burden, as the trials in this review provided no information on the impact of treatment on any of these outcomes. It will also be useful for trials to further investigate the effectiveness of psychological interventions, and for studies to recruit participants with anxiety only, as well as those with co‐morbid anxiety and depression. Research into the uptake and acceptability of psychological interventions for anxiety after stroke would also be valuable.

Summary of findings

Open in table viewer
Summary of findings for the main comparison. Interventions for treating anxiety after stroke

Interventions for treating anxiety after stroke

Patient or population: stroke survivors with anxiety
Settings: out of hospital
Intervention: pharmacological or psychological treatment

Outcomes

Illustrative comparative risks* (95% CI)

Relative effect
(95% CI)

No. of Participants
(studies)

Quality of the evidence
(GRADE)

Comments

Assumed risk

Corresponding risk

Control

Interventions

Proportion of stroke patients without clinical diagnosis of an anxiety disorder

See comment

See comment

Not estimable

19
(1 study)

⊕⊝⊝⊝
very lowa,b

Clinical anxiety at 3 months: 4/9 in intervention group no longer had anxiety, 1/10 in control group no longer had anxiety

Proportion of stroke patients scoring outside anxiety range; or changes from baseline on an anxiety rating scale

See comment

See comment

Not estimable

196
(3 studies)

⊕⊝⊝⊝
very lowc,d

Statistically significant difference in anxiety scores on HADS‐A scale at 3 months, with reduction in anxiety for those using therapeutic CD (P value = 0.001); statistically significant differences in HAM‐A scores at 6 weeks and 4 weeks with reduced anxiety for those taking paroxetine and paroxetine with psychological therapy and those taking buspirone, respectively (P value < 0.01)

Co‐morbid depression

See comment

See comment

Not estimable

175
(2 studies)

See comment

Reduction in depression symptoms according to HAM‐D at 6 weeks and at 4 weeks for those taking paroxetine and paroxetine with psychological therapy and those taking buspirone, respectively

Quality of life ‐ not reported

See comment

See comment

Not estimable

See comment

Outcome not reported in any study

Social activities ‐ not reported

See comment

See comment

Not estimable

See comment

Outcome not reported in any study

Activities of daily living

See comment

See comment

Not estimable

81
(1 study)

⊕⊝⊝⊝
very lowb,e

Improvement in activities of daily living in all groups, but greatest improvement in those taking paroxetine with psychological therapy

Principal caregiver burden ‐ not reported

See comment

See comment

Not estimable

See comment

Outcome not reported in any study

*The basis for the assumed risk (e.g. median control group risk across studies) is provided in footnotes. The corresponding risk (and its 95% confidence interval) is based on the assumed risk in the comparison group and the relative effect of the intervention (and its 95% CI)
CI: confidence interval; RR: risk ratio

GRADE Working Group grades of evidence
High quality: Further research is very unlikely to change our confidence in the estimate of effect
Moderate quality: Further research is likely to have an important impact on our confidence in the estimate of effect and may change the estimate
Low quality: Further research is very likely to have an important impact on our confidence in the estimate of effect and is likely to change the estimate
Very low quality: We are very uncertain about the estimate

aStudy was unblinded, with further risks of recruitment bias and drop‐outs. Downgraded one level for risk of bias
bOnly one study with small number of participants, downgraded two levels for imprecision
cLimited detail in Wang 2005 and Zhang 2005 for effective assessment of bias; lack of blinding, risks of recruitment bias, and drop‐outs in Golding 2016. Downgraded two levels for risk of bias
dOnly three studies with few participants, all with different interventions that are not comparable. Downgraded one level for indirectness and one level for imprecision
eLimited detail in studies, unable to effectively assess risk of bias; downgraded one level

Background

Description of the condition

Stroke and anxiety disorders are major public health problems. Although stroke is the leading cause of adult disability (Department of Health 2007; Intercollegiate Stroke Working Party 2012), anxiety is the most common mental health disorder (Lepine 2002). Prevalence of anxiety after stroke ranges from 20% to 25% (Campbell Burton 2013), and it remains a common problem several years after the stroke event (Ayerbe 2013; Langhorne 2000). Anxiety is more common among younger or female people after stroke, those unable to work after stroke, and those from lower income backgrounds (Ayerbe 2013; Broomfield 2015; Menlove 2015). 

Several distinct types of anxiety disorders are known, such as general anxiety disorder (GAD), panic disorder, social phobia, obsessive‐compulsive disorder (OCD), and post‐traumatic stress disorder (PTSD). Although categorically different, these disorders share similar hallmark characteristics of excessive and irrational fear, subjective apprehension, and difficulty and distress in managing daily tasks (Gelder 2006). Furthermore, although an anxiety disorder is diagnosed, many individuals experience significant levels of physical (e.g. heart palpitations, shortness of breath), cognitive (e.g. feeling of losing control), or behavioural (e.g. avoidance of certain stimuli) symptoms of anxiety that can affect their daily lives. All types of anxiety disorders have been observed in stroke patients (House 1991; Max 2002), and have been shown to have a negative impact on quality of life (Ahlsio 1984). Co‐morbidity with depression is also very high (Castillo 1993). Studies have found that depression is more severe and longer lasting in those with co‐morbid anxiety (Shimoda 1998), and stroke patients with co‐morbid anxiety and depression have higher levels of impairment in activities of daily living, greater cognitive impairment, and fewer social ties than those with depression alone (Shimoda 1998).

Differentiating between normal worries and emergence of pathological anxiety disorders, or clinically significant levels of anxiety symptoms, is difficult for several reasons. Advanced age and limited verbal ability, both of which are common within the stroke population, increase the difficulty involved in identifying persons with anxiety (Van Rijswijk 2009). Other practical problems, such as difficulty accessing specialist mental health services, not presenting for treatment, and lack of clinical guidelines specific to stroke patients with anxiety problems, mean that individuals may go untreated (Fernandez 2007).

Description of the intervention

We were interested in pharmaceutical, psychological, or any alternative therapy whose primary purpose was to treat anxiety disorders or significant levels of anxiety symptoms in stroke patients. Given the potential diversity of anxiety states, we did not limit our criteria to an a priori list of therapies. However, we did expect to find studies that treated anxiety according to evidence‐based guidelines, such as those recommended by the National Institute for Health and Clinical Excellence (NICE 2011), which outline pharmaceutical and psychological interventions that can be used to treat members of the general population with specific anxiety disorders. To our knowledge, no specific guidelines have been developed for the treatment stroke patients with anxiety.

Pharmaceutical therapies

Several classes of drugs can be used to treat anxiety disorders. These drugs vary according to the neurotransmitters that they are purported to affect.

Selective serotonin reuptake inhibitors (SSRIs) are a class of antidepressant drugs used to treat anxiety. Serotonin is a neurotransmitter involved in regulating mood. SSRIs, such as fluoxetine, sertraline, escitalopram, paroxetine, and citalopram, are commonly prescribed for panic disorder, OCD, PTSD, and social phobia (NIMH 2009). Pharmacologically, SSRIs inhibit post‐release reuptake of serotonin by presynaptic nerve terminals, hence increasing the level of available serotonin in the brain (Craig 2003). 

Tricyclic antidepressants (TCAs) (e.g. imipramine) are an older generation of antidepressant drugs developed in the 1950s; they have been replaced for the most part by SSRIs. However, TCAs are still recommended in clinical guidelines for treating GAD and panic disorder (NICE 2011). TCAs act as serotonin and norepinephrine reuptake inhibitors, which results in increased extracellular concentrations of these neurotransmitters and hence enhanced neurotransmission.

Benzodiazepines (e.g. diazepam, alprazolam) are anxiolytics used to treat GAD and social phobia (Baldwin 2005), and in some instances specific phobia (NICE 2011; NICE 2014). These drugs enhance the effect of the gamma‐aminobutyric acid (GABA) neurotransmitter, thereby reducing the somatic symptoms associated with anxiety, such as muscle tension and insomnia, but they are recommended only for short‐term use.

Zopiclone, zolpidem, and zaleplon (Z‐drugs) are hypnotics that can be prescribed to help patients with the sleep disturbance associated with GAD and PTSD (NICE 2005). These drugs behave in a similar way to benzodiazepines, except they have a shorter half‐life. 

Psychological therapies

Various forms of psychological therapies are available for treating anxiety. They are particularly suited to certain forms of anxiety, such as social anxiety (NICE 2013), and may be welcomed by individuals (especially older people) who may prefer not to use psychotropic drugs. This preference is based on concern about dependence, prior negative experiences, and the fact that many individuals do not view their psychological symptoms as a medical illness. Several forms of psychological therapies are described below.

Behaviour therapy is based on learning theory and consists of approaches for developing adaptive ways of behaving. The aim of behaviour therapy is to treat anxiety through techniques designed to reinforce desired behaviours while eliminating undesired behaviours. 

Cognitive therapy is based on the cognitive model, which hypothesises that a person’s emotions and behaviours are influenced by their perception of events. Hence it is not the situation itself that determines how people feel but rather the way they construe the situation (Beck 1979).

Cognitive‐behavioural therapy (CBT) incorporates elements from both cognitive and behavioural therapies with the goal of changing a person’s thoughts, beliefs, attitudes, and expectations, and how a person acts (similar to behaviour therapy). It is 'present‐centred' and directs individuals to identify the current issues that are causing them distress, with the support of a trained psychological practitioner. Individuals talk with their therapist about specific problems in a structured manner and may be given homework consisting of activities to be completed before their next session. CBT is characterised as structured, goal‐oriented, and time‐limited (Beck 1997).

Complementary or alternative therapies

Although we cannot provide an exhaustive description of all interventions that can be used to treat anxiety, patients may choose from a mix of alternative therapies. For example, self‐help manuals may assist patients in gaining understanding or insight into their emotional problems and can be used to treat anxiety disorders or severe symptoms with limited therapist involvement (Van Boeijen 2005). Other therapies such as exercise training, which may act as a buffer for stress or trigger the release of monoamine neurotransmitters, and relaxation therapy, which teaches individuals to recognise symptoms of anxiety and how to respond to them via a technique that reduces arousal, have also been used to treat anxiety (Jorm 2004).

How the intervention might work

Pharamaceutical interventions work by altering the level of certain neurotransmitters in the brain, and psychological interventions aim to alter maladaptive behaviour and cognition to improve emotional functioning. Treatments in the complementary and alternative category work through multiple mechanisms. Additionally, patients receiving standard care, or those waiting to receive an intervention, may experience a reduction in anxiety symptoms through a placebo effect that is not directly related to the action of the intervention or treatment.

Why it is important to do this review

Anxiety after stroke has received substantially less attention than other psychological outcomes by both clinicians and researchers. Systematic reviews have been carried out to assess the effectiveness of interventions used to treat depression and emotionalism when they occur after stroke (Hackett 2008; Hackett 2010). The previous version of this Cochrane Review included only two small trials, thus highlighting a gap in the literature and knowledge base (Campbell Burton 2011). Studies in stroke (Shimoda 1998) and non‐stroke populations (Wittchen 2003) have shown that anxiety increases the risk and severity of depression. Hence, early treatment of anxiety may reduce the risk of subsequent depression and its associated adverse consequences. Clinical guidelines for treating anxiety have been established, but their effectiveness in stroke populations remains unknown. We chose to evaluate any intervention whose primary aim was to treat anxiety after stroke, as evidence suggests diversity among patient preferences (Hyde 2005; Riedel‐Heller 2005). We did expect that most of the trials retrieved would provide pharmaceutical or psychologically based interventions.

Objectives

The primary objective was to assess the effectiveness of pharmaceutical, psychological, complementary, or alternative therapeutic interventions in treating stroke patients with anxiety disorders or symptoms. The secondary aim was to identify whether any of these interventions for anxiety had an effect on quality of life, disability, depression, social participation, caregiver burden, or risk of death.

Methods

Criteria for considering studies for this review

Types of studies

Randomised controlled trials (RCTs) in which the primary aim of the intervention was to treat anxiety in people with a clinical diagnosis of stroke (Hatano 1976) were eligible for inclusion in this review. Review authors applied no restrictions on language or study location. We expected eligible trials to compare the effect of an intervention plus usual care against placebo, a different intervention, or different doses or frequency of interventions. Trials had to include a placebo or standard care control arm; otherwise they were not eligible for inclusion.

Types of participants

All stroke patients enrolled into an RCT must have received a clinical diagnosis of an anxiety disorder according to the Diagnostic and Statistical Manual of Mental Disorders (DSM‐III (APA 1980), DSM‐III‐R (APA 1987), DSM‐IV (APA 1994), DSM‐IV‐TR (APA 2000)) or had to meet similar diagnostic criteria. Stroke patients in RCTs deemed to have significant levels of anxiety symptoms as established by a predetermined defined cut‐off score on an anxiety screening tool were also eligible. Review authors applied no restrictions on age distribution or gender. Studies with mixed populations of ischaemic or haemorrhagic stroke were eligible, but we excluded studies assessing treatment effect in an exclusively subarachnoid haemorrhage patient population, as the characteristics, treatment, and management of these patients can be substantially different from those of other stroke patients. Studies treating stroke patients for other conditions such as depression, cognitive impairment, or physical disability were also ineligible, unless we could determine that all patients had co‐morbid anxiety upon enrolment into the trial and that treatment for anxiety was one of the main objectives of the trial.      

Types of interventions

We evaluated RCTs comparing pharmaceutical interventions administered to stroke patients versus placebo or standard care. The drug had to be administered for the purpose of treating anxiety. We excluded trials in which drugs were administered for other purposes, such as neuroprotection. We also evaluated RCTs that compared psychological interventions versus placebo or standard care for the purpose of treating anxiety. We expected that these types of interventions would have a clearly defined psychological component; would be structured, delivered, and supervised by trained staff; and would be time‐limited. We excluded interventions whose purpose was to prevent anxiety or simply to provide information or educate patients. We did not include trials of interventions such as occupational therapy or co‐ordinator visitation for stroke support unless they had a definitive psychological component aimed at treating anxiety.

Types of outcome measures

Primary outcomes

  1. Proportion of stroke patients without a clinical diagnosis of an anxiety disorder according to the DSM (APA 1994) or another standard diagnostic classification at the end of scheduled follow‐up

  2. Proportion of stroke patients scoring outside the anxiety symptom range (as defined by study authors); or with changed scores from baseline on an anxiety rating scale or via self‐report at the end of scheduled follow‐up  

Secondary outcomes

  1. Co‐morbid depression, as diagnosed by  DSM or determined by a depression rating scale such as the Beck Depression Inventory (BDI) (Beck 1961), the Hamilton Depression Scale (HAM‐D) (Hamilton 1960), or the Montgomery‐Asberg Depression Rating Scale (Montgomery 1979)

  2. Quality of life as measured on scales such as the 36‐item Short Form Questionnaire (SF‐36) (Ware 1993)

  3. Social activities as measured on scales such as the Frenchay Activities Index (Wade 1985)

  4. Activities of daily living as measured on scales such as the Barthel Index (Mahoney 1965)

  5. Principal caregiver burden as measured by scales such as the Zarit Caregiver Burden Interview (Zarit 1980)

  6. Any adverse consequence resulting from treatment for anxiety such as drug tolerance, co‐dependence on the counsellor, or death. We also recorded rates of loss to follow‐up in different arms of trials as a possible indicator of treatment acceptability 

Search methods for identification of studies

See the 'Specialized Register' section of the Cochrane Stroke Group module. We attempted to identify all relevant trials regardless of language or publication status, and we arranged translation of relevant papers when necessary.

Electronic searches

We searched the trials register of the Cochrane Stroke Group (January 2017), the Cochrane Central Register of Controlled Trials (CENTRAL; the Cochrane Library; 2017, Issue 1: searched January 2017) (Appendix 1); MEDLINE (1966 to January 2017) in Ovid (Appendix 2); Embase (1980 to January 2017) in Ovid (Appendix 3); the Cumulative Index to Nursing and Allied Health Literature (CINAHL; 1937 to January 2017) in EBSCO (Appendix 4); and PsycINFO (1800 to January 2017) in Ovid (Appendix 5).

We developed the MEDLINE search strategy with the help of the Cochrane Stroke Group Information Specialist and adapted it for use with the other databases. We searched the following trial registers to identify additional published, unpublished, and ongoing clinical trials.

  1. ClinicalTrials.gov (http://clinicaltrials.gov) (January 2016).

  2. World Health Organization (WHO) International Clinical Trials Registry Portal (http://apps.who.int/trialsearch/) (September 2016).

  3. ISRCTN Registry (http://www.isrctn.com/) (January 2016).

Searching other resources

We identified reviews from the results of database searches and conducted backward citation searches for potentially eligible trials. We used Google Scholar (scholar.google.co.uk) to conduct forward citation searching of included studies. We contacted known researchers to ask for information on completed and ongoing clinical trials.

Data collection and analysis

Selection of studies

Two review authors (SRL, H‐YYC) independently screened all reports yielded by the searches of electronic databases, and excluded citations that were clearly irrelevant based on title and abstract. We retrieved the full texts of remaining articles and reviewed them for inclusion on the basis of eligibility criteria for the review. If consensus could not be reached, we consulted a third review author (PK) for adjudication.

Data extraction and management

Two review authors (SRL and H‐YYC) independently extracted data and recorded them on a paper extraction form designed to capture key information. The two review authors reconciled the data extraction and entered the data into Review Manager 5 (RevMan 2014). We recorded core data elements such as study details, methods, information about participants, and outcomes for analysis.

Assessment of risk of bias in included studies

We assessed study bias in accordance with the Cochrane tool for assessing risk of bias (Higgins 2011). This instrument includes six domains whereby different types of potential bias can be evaluated, including sequence generation, allocation concealment, blinding (of participants, personnel, and outcome assessors), incomplete outcome data, selective outcome reporting, and other unspecified types of bias (e.g. conflict of interest). We identified respective biases from each study and displayed them in a tabular format. We summarised risks qualitatively and attempted to describe their impact on research findings. 

Measures of treatment effect

We prepared a narrative description of all studies. Included trials measured anxiety using the Hamilton Anxiety Scale (HAM‐A) (Hamilton 1959) and the Hospital Anxiety and Depression Scale (HADS). The HAM‐A, a rating scale that was developed to quantify the severity of anxiety symptoms, is often used in psychotropic drug evaluation. It consists of 14 items, each defined by a series of symptoms. Each item is rated on a five‐point scale, ranging from 0 (not present) to 4 (severe). Total scores on the HAM‐A range from 0 to 56. A score of 14 or higher has been suggested to indicate clinically significant anxiety (Maier 1988). The HADS is commonly used to assess levels of patient anxiety and depression. The HADS evaluates 14 items (seven for anxiety and seven for depression) and uses a scale of 0 to 3 for each item, with a total score of 21 possible for each subscale (Zigmond 1983). Scores of 8 or above on either HADS subscale are commonly taken to indicate clinical significance (Bjelland 2002).

Unit of analysis issues

In the event that outcomes were repeatedly observed in participants (e.g. follow‐up at four and six weeks), we reported the measurement taken at the longest time point post intervention from each study.

Dealing with missing data

We planned to contact study authors to obtain information about missing data and, if we could not obtain this, we planned to conduct a 'what if' sensitivity analysis to explore the impact that missing data could have on the final outcome.

Assessment of heterogeneity

The intent was to measure heterogeneity by using the I2 statistic. If higher than 50% (a level considered moderate to substantial), we would have calculated the treatment effect by using the random‐effects method, which assumes that different studies are estimating different but related intervention effects and so provides a more conservative intervention effect estimate and wider confidence intervals (DerSimonian 1986).

Assessment of reporting biases

We planned to construct a funnel plot estimate to assess the potential influence of reporting bias if we had included more than 10 studies in the systematic review.

Data synthesis

Two review authors (SRL and H‐YYC) independently extracted data from the included studies. One review author (SRL) entered data into RevMan (RevMan 2014) and the other (H‐YYC) cross‐checked the data entered. Review authors resolved disagreements by referring to the original study report.

Subgroup analysis and investigation of heterogeneity

Several factors could impact study heterogeneity and effect size. We initially planned to undertake subgroup analyses on certain clinically relevant factors, such as specific type of anxiety disorder (e.g. GAD, social phobia), length of time treatment was administered, or length of time since stroke at entry into the trial.

Sensitivity analysis

To test robustness of findings and examine the degree to which findings influenced effect size, we planned to analyse data and include studies that executed allocation concealment, double blinding, and fidelity to administered intervention to the highest standard.

'Summary of findings' table

We used the principles of GRADE (Grades of Recommendation, Assessment, Development and Evaluation Working Group; Guyatt 2008) to assess the quality of the body of evidence associated with the following specific outcomes in our review.

  1. Proportion of patients without a clinical diagnosis of anxiety.

  2. Proportion of patients scoring outside the anxiety symptom range; or with change scores from baseline on an anxiety rating scale.

  3. Co‐morbid depression.

  4. Quality of life.

  5. Social activities.

  6. Activities of daily living.

  7. Principal caregiver burden.

We constructed a 'Summary of findings' table by using GRADE software (gradepro.org). The GRADE approach appraises the quality of the body of evidence according to the extent to which one can be confident that an estimate of effect or association reflects the item assessed. The quality of a body of evidence is based on within‐study risk of bias (methodological quality), directness of the evidence, heterogeneity of the data, precision of effect estimates, and risk of publication bias.

Results

Description of studies

We identified no trials that compared any intervention with a placebo control. See Characteristics of included studies and Characteristics of excluded studies.

Results of the search

We identified 4223 records for the 2017 update through electronic database searches. We contacted three researchers known to be working in the field, who provided information on their current work to enable us to assess study eligibility. We identified 20 additional records from the Cochrane Stroke Group trials register. We removed duplicates and sifted 4069 titles and abstracts. We identified nine reviews for backward citation searching and evaluated potential studies from these reviews, alongside 39 potentially relevant full‐text articles; we identified one study that met the inclusion criteria for this review (Golding 2016), in addition to the two studies in the previous review update (Wang 2005; Zhang 2005). See Figure 1.


Search flow diagram (from searches conducted for the review update, October 2010 to January 2017).

Search flow diagram (from searches conducted for the review update, October 2010 to January 2017).

Included studies

Three trials with a total of 196 randomised participants met our inclusion criteria (Golding 2016; Wang 2005; Zhang 2005).

Golding 2016 conducted what is described as a 'pilot study' to assess the effectiveness of a self‐help relaxation therapy. Participants were 21 stroke survivors who had anxiety and were living in the community. After a telephone interview and assessment, intervention group members (11 participants) were sent a self‐help autogenic relaxation CD; they were asked to follow the instructions on it five times per week for one month; and they were asked to complete a diary sheet. The control group (10 participants) did not receive the CD until the end of the study period at three months. Individuals were excluded if they were unable to complete rating scales via telephone, had a Telephone Interview Cognitive Score (TICS) ≤ 20, had significant difficulties with language or were non‐English speaking, had a co‐morbid psychiatric disorder other than an affective disorder, or were currently receiving other psychological interventions. Study investigators determined anxiety at baseline and at one month, two months, and three months using the HADS anxiety subscale (HADS‐A). They did not measure any additional outcomes. The mean age of participants was 67.8 years in the intervention group, and 62.4 years in the control group.

Wang 2005 evaluated the effectiveness of the SSRI paroxetine and of combination paroxetine and psychotherapy. Eighty‐one first‐ever stroke patients who met Chinese Classification and Diagnostic Criteria of Mental Disorders (CCMD‐3) criteria were randomised to one of the three groups. The first group (27 participants) received 20 mg of paroxetine per day, and the second group (27 participants) received the same amount of paroxetine per day along with psychiatrist‐administered supportive psychotherapy for 30 to 60 minutes once per week. A parallel control group with 27 participants received routine treatment only. Study authors did not specify the length of time since stroke at the time of participant recruitment. Patients who were in a coma or aphasic, had severe cognitive dysfunction or other serious disease, or who had been prescribed depression or antipsychotic medications in the three months before the start of the trial were excluded. Investigators provided interventions for six weeks and used HAM‐A and HAM‐D scales to assess the severity of anxiety and depression symptoms at baseline and at two, four, and six weeks during treatment. They assessed scores on the Barthel Index measuring activities of daily living at all time points. The mean age of participants was 62.4 years in the drug only group, 64.0 years in the drug plus psychotherapy group, and 63.2 years in the standard care group.

Zhang 2005 examined the effect of the anxiolytic drug buspirone hydrochloride against standard care. Researchers recruited 94 stroke patients with co‐morbid anxiety and depression according to the CCMD‐3. They deemed that individuals in an unstable condition were ineligible but provided no description of the unstable conditions. Investigators administered buspirone for four weeks to those in the intervention arm of the study at 20 to 30 mg per dose during the first week and at 40 to 60 mg per dose during the second week. They provided no information about the amount administered during the third or fourth week. Researchers measured anxiety and depression using HAM‐A and HAM‐D scales at baseline, and at two and four weeks during the intervention. The mean age of participants was 57.8 years for the intervention group and 59.2 years for the control group. Study authors reported no other secondary outcomes of interest.

Excluded studies

We excluded 40 studies after assessing the full text of the article during the most recent search. We excluded 24 of these studies as they used the wrong study design, did not include participants who had a diagnosis of stroke, or did not include a treatment aimed at reducing anxiety. Sixteen studies measured anxiety, often alongside depression, and aimed to relieve psychological symptoms exclusively or in addition to physical symptoms. None of these studies required participants to have a clinical diagnosis of anxiety for study participation (Aidar 2012; Aidar 2013; Akerlund 2013; Chaiyawat 2012; Chan 2012; Hoffmann 2015; Ihle‐Hansen 2014; Immink 2014; Jouzi 2010; Karaiskos 2012; Kongkasuwan 2014; Kulishova 2014; Mikami 2014; Peng 2015; Wu 2012; Xue 2013). We excluded these key studies and listed them under Characteristics of excluded studies. In addition, we excluded eight key trials from the original review (Kimura 2003; Li 2005; Liu 2004; Mok 2004; Morrison 1998; Rorsman 2006; Wu 2008; Ye 2006) and reported reasons for exclusion under Characteristics of excluded studies.

Studies awaiting classification

Four studies are awaiting classification (Doogan 2012; Guilan 2013; Kerr 2014; Yates 2015). These studies were published as abstracts only, without author contact details. All studies included participants who were stroke survivors but provided insufficient detail to establish whether included participants were required to have a diagnosis of anxiety.

Ongoing studies

We identified no eligible studies in clinical trials registers.

Risk of bias in included studies

We have provided a summary of 'Risk of bias' assessments in Figure 2.


Risk of bias summary: review authors' judgements about each risk of bias item for each included study.

Risk of bias summary: review authors' judgements about each risk of bias item for each included study.

Allocation

Golding 2016 used a random number generator that we judged to be adequate with low risk of bias. However, details in the paper were insufficient to show whether allocation of participants was adequately concealed. Wang 2005 stated that investigators used simple random sampling, and Zhang 2005 indicated researchers used a random number list for participants who met the inclusion criteria. However, neither study described the randomisation process, hence the integrity of sequence generation and allocation concealment was unclear.

Blinding

It was not possible to blind participants to the study intervention in Golding 2016, which presented an inevitably high risk of performance and detection bias. Neither Wang 2005 nor Zhang 2005 provided information about blinding. As these studies included no placebo control group, blinding would likely be possible only for independent outcome assessors.

Incomplete outcome data

Golding 2016 was a small study with few but potentially significant losses and no intention‐to‐treat analysis; therefore, we judged this trial to have high risk of attrition bias. Wang 2005 reported no loss to follow‐up and did not describe adherence to the treatment protocol. Zhang 2005 reported outcomes for participants who remained until study completion. Hence, this study is classified as an 'available case analysis'.

Selective reporting

We found no evidence of selective outcome reporting in any of the included trials. Investigators reported all outcomes as described in the methods section of the full study report. However, we did not obtain the research protocols, and study authors did not report clinical trial registration, so we do not know if other outcomes were measured but not reported.

Other bias

Methods of recruitment in Golding 2016 had increased risk of bias, as interested participants responded to advertisements in publications intended for stroke survivors. It is possible that participants who contacted the research team had a positive bias towards psychological therapies for treatment of anxiety, although the study did not assess this potential bias.

Effects of interventions

See: Summary of findings for the main comparison Interventions for treating anxiety after stroke

It was not appropriate to combine trial data on pharmacological therapies with data on relaxation therapy; therefore we did not perform a meta‐analysis.

In summary, Golding 2016 found preliminary evidence in this pilot study that an autogenic relaxation CD may reduce anxiety among stroke survivors living in the community. Wang 2005 found that both paroxetine and paroxetine plus psychotherapy reduced the severity of anxiety symptoms as measured by the HAM‐A when compared with standard care. Zhang 2005 found that buspirone hydrocholoride was effective in reducing anxiety symptoms when compared with standard care.

We have described the effectiveness of interventions compared with standard care in the prespecified outcomes below and have reported results of the GRADE assessment, with explanations of decisions for each outcome, in summary of findings Table for the main comparison.

Primary outcomes

Proportion of stroke patients without a clinical diagnosis of an anxiety disorder

In Golding 2016, four members of the intervention group were no longer considered to have clinical levels of anxiety at three months, compared with one participant in the control group. Investigators reported loss of two participants after randomisation in the intervention group with no intention‐to‐treat analysis. Therefore, results include nine participants in the intervention group and 10 participants in the control group. We judged this evidence as having very low quality.

Wang 2005 and Zhang 2005 did not report this outcome.

Proportion of stroke patients scoring outside the anxiety symptom range; or with changed scores from baseline on an anxiety rating scale

Golding 2016 measured levels of anxiety at one, two, and three months post stroke, but, as per the review protocol, we have considered analysis only at the final time point. At three months post stroke, mean scores on the HADS‐A scale were 6.9 (± standard deviation (SD) 4.9) in the intervention group, and 11.0 (SD ± 3.9) in the control group, showing a statistically significant difference (P value = 0.001). Again these results include nine participants in the intervention group and 10 participants in the control group.

In Wang 2005, mean HAM‐A anxiety scores at baseline in the drug only, drug plus psychotherapy, and standard care groups were 14.0 (SD ± 2.8), 13.9 (SD ± 2.9), and 13.8 (SD ± 2.8), respectively. At six weeks, mean anxiety scores were significantly lower in the two intervention groups relative to the controls at 5.4 (SD ± 1.7) and 3.8 (SD ± 1.8) in the drug only and drug plus psychotherapy groups, but the mean anxiety score was 12.8 (SD ± 1.9) in the control group. Relative to the standard care group, this represents 58% and 71% lower mean anxiety scores in the paroxetine and paroxetine plus psychotherapy groups, respectively. Cohen's d was 4.10 for the paroxetine only group versus the control group, and 4.86 for the paroxetine plus psychotherapy group versus the control group. Both of these differences were statistically significant (P value < 0.01).

In Zhang 2005, four weeks after trial initiation, the mean anxiety score on the HAM‐A decreased from 22.7 (SD ± 5.2) to 6.5 (SD ± 3.1) in the intervention group. This decrease was significantly larger than that seen in the standard care group (P value < 0.01), for which the mean anxiety score decreased from 22.5 (SD ± 4.3) to 12.6 (SD ± 3.4) after four weeks. The mean anxiety score in the intervention group was 50% lower than that in the standard care group (Cohen's d effect size = 1.87).

HAM‐A scores range from zero to 56; a score greater than 14 indicates mild to moderate anxiety symptoms. Study authors in Golding 2016 used a lower cut‐off of ≥ 6, which they recommended as the most sensitive for a stroke population. On this basis, the reduction in anxiety scores among intervention groups in each trial appears to be clinically meaningful. However, by using GRADE, we judged that all evidence for this outcome was of very low quality.

Secondary outcomes

Co‐morbid depression

The possible range on the HAM‐D is zero to 54, with higher scores indicating more severe symptoms. In Wang 2005, mean depression severity scores were 18.2 (SD ± 1.4), 18.8 (SD ± 3.1), and 18.0 (SD ± 1.3) at baseline in the paroxetine, paroxetine plus psychotherapy, and standard care groups, respectively. Although results showed no change in the control group after six weeks (mean 17.5, SD ± 1.1), both the drug only group and the drug plus psychotherapy group had significantly fewer depression symptoms (mean 10.1, SD ± 1.1; mean 8.9, SD ± 1.2), respectively.

In Zhang 2005, buspirone was effective in significantly reducing depression symptoms as measured on the HAM‐D in the intervention group compared with the control group. The mean depression score decreased from 24.6 (SD ± 4.7) to 8.3 (SD ± 2.8) in the intervention group, and from 23.4 (SD ± 5.3) to 13.4 (SD ± 2.7) in the standard care group.

We judged that the evidence for this outcome was of very low quality.

Quality of life

Studies did not report this outcome.

Social activities

Studies did not report this outcome.

Activities of daily living

Only one trial reported changes in functional status as measured by the Barthel Index of activities of daily living (ADLs) (Wang 2005). Investigators found that ADLs improved significantly in all three groups of participants, with the greatest improvement noted in the drug plus psychotherapy group (which increased from 62.0 (SD ± 23.1) to 90.2 (SD ± 7.3)), followed by the drug‐only group (which increased from 60.9 (SD ± 23.9) to 84.3 (SD ± 8.4)), with standard care controls showing the least improvement (the increase was from 61.5 (SD ± 24.3) to 78.3 (SD ± 15.0)). We judged the evidence for this outcome to be of very low quality.

Principal caregiver burden

Studies did not report this outcome.

Adverse consequences

Wang 2005 reported 26 adverse events, all in participants given paroxetine or paroxetine with psychotherapy; nine participants given paroxetine reported nausea and vomiting, and five reported dizziness, 10 participants given paroxetine with psychotherapy reported nausea and vomiting, and two reported dizziness.

In Zhang 2005, three participants reported dizziness and two reported palpitations. Again, all adverse events occurred in the intervention group.

Combining data from Wang 2005 and Zhang 2005 revealed that intervention agents increased the risk of dizziness (risk ratio, Mantel‐Haenszel, random‐effects 7.32 (95% confidence interval 0.96 to 55.95)). See Analysis 1.1.

In Golding 2016, one participant reported that the training made his "eyes feel funny", and participants did not describe any other adverse consequences of therapy.

Loss to follow‐up and intervention fidelity

No participants were lost to follow‐up in Wang 2005. However, in both intervention and control groups, 23% of participants were lost in Zhang 2005. Reasons given for drop‐out in the intervention group were unsatisfactory treatment effect, drug side effects, and subsequent prescription of benzodiazepines. Recurrent stroke, prescribing of benzodiazepines, and withdrawal were reasons given for loss to follow‐up in the control group. Wang 2005 and Zhang 2005 did not report data on intervention fidelity (other than loss to follow‐up). Loss of two participants in the intervention group in Golding 2016 was due to personal reasons and to a change in health circumstances; two of the participants in the intervention group used the CD less than once a week, rather than five times per week as directed.

Discussion

Summary of main results

We found three published trials and were unable to identify any ongoing trials. Among the three published trials, anxiety symptom severity as measured by the Hamilton Anxiety Scale (HAM‐A) or the Hospital Anxiety and Depression Scale (HADS) was the outcome of interest. None of these studies evaluated clinical anxiety disorders or included a placebo control group. Study results suggest that both paroxetine and buspirone are effective pharmacological therapies for treatment of anxiety after stroke. However, in the absence of a placebo control arm, the true level of effectiveness is unknown. Combining paroxetine and psychotherapy did not confer significant additional benefit for stroke patients. Paroxetine appeared to be well tolerated, as no drop‐outs occurred among participants, but a large proportion experienced symptoms of nausea or dizziness. Buspirone was also effective in reducing anxiety, but investigators reported substantial loss to follow‐up and some adverse events. Loss to follow‐up in the buspirone trial is unusual as results show an equally high level of drop‐out in the control group. The addition of Golding 2016 to the most recent update provides limited evidence that relaxation therapy may reduce anxiety among stroke survivors.

Overall completeness and applicability of evidence

This review was intentionally broad because we suspected that the literature on interventions used to treat anxiety after stroke was not as established as for some of the other post‐stroke psychological conditions. We attempted to collate comprehensive evidence relevant to the review question by conducting a thorough evidence search.

In the original review, the two included studies provided very little information about the populations from which participants were selected, and we could not ascertain whether the results were generalisable to the stroke population (Campbell Burton 2011). We also noted that the inclusion criteria for the two trials included in the original review required participants to have both anxiety and depression according to the Chinese Classification and Diagnostic Criteria of Mental Disorders (CCMD‐3), and therefore, the evidence could not be attributable to stroke survivors with anxiety alone. The inclusion of Golding 2016 in this most recent update provided evidence for a more specific community‐based population of stroke survivors with anxiety; however, study authors used a much lower threshold for clinical anxiety with the HAM‐A scale than in previous studies. Indeed, it should be noted that although the HAM‐A is widely used in pharmaceutical studies of anxiety, it is not appropriate for use as a diagnostic or screening instrument. The HAM‐A focuses primarily on the phobic and autonomic arousal symptoms of anxiety, and gives little weight to the psychic symptoms. Given the physical consequences of stroke, it would be misleading to attribute all physical symptoms solely to anxiety after stroke. Therefore, the evidence presented in this review is limited by the measurement scales used to assess anxiety in this population.

Quality of the evidence

Two studies assessing pharmacological agents provided limited methodological details for adequate judgement of risk of bias across all domains. The third study, which assessed relaxation therapy, inevitably had high risk of bias due to the inability to blind participants to treatment allocation. Clinical trial registration was lacking in all studies and study sample sizes were small, including one study with just 21 participants, which reported two drop‐outs and had high risk of recruitment bias. The pharmacological studies inadequately described comparison groups. In using GRADE to assess the quality of the evidence, we were particularly concerned about risk of bias in these studies, as well as the limited number of studies including few participants. We downgraded the evidence by two levels for risk of bias and by one or two levels for imprecision; we therefore rated the evidence for each reported outcome in this review as very low.

Potential biases in the review process

To the extent possible, we worked to ensure minimal bias in the review process. We undertook an extensive literature search guided by the Cochrane Stroke Group, and we contacted key researchers in the field to obtain information about studies with a focus on post‐stroke anxiety. Additionally, we did not limit findings to English language papers. Two review authors independently decided whether studies should be included and independently extracted data.

Agreements and disagreements with other studies or reviews

To our knowledge, no other systematic reviews have examined interventions used to treat anxiety after stroke.

Search flow diagram (from searches conducted for the review update, October 2010 to January 2017).
Figures and Tables -
Figure 1

Search flow diagram (from searches conducted for the review update, October 2010 to January 2017).

Risk of bias summary: review authors' judgements about each risk of bias item for each included study.
Figures and Tables -
Figure 2

Risk of bias summary: review authors' judgements about each risk of bias item for each included study.

Comparison 1 Pharmacological agents versus control, Outcome 1 Dizziness.
Figures and Tables -
Analysis 1.1

Comparison 1 Pharmacological agents versus control, Outcome 1 Dizziness.

Summary of findings for the main comparison. Interventions for treating anxiety after stroke

Interventions for treating anxiety after stroke

Patient or population: stroke survivors with anxiety
Settings: out of hospital
Intervention: pharmacological or psychological treatment

Outcomes

Illustrative comparative risks* (95% CI)

Relative effect
(95% CI)

No. of Participants
(studies)

Quality of the evidence
(GRADE)

Comments

Assumed risk

Corresponding risk

Control

Interventions

Proportion of stroke patients without clinical diagnosis of an anxiety disorder

See comment

See comment

Not estimable

19
(1 study)

⊕⊝⊝⊝
very lowa,b

Clinical anxiety at 3 months: 4/9 in intervention group no longer had anxiety, 1/10 in control group no longer had anxiety

Proportion of stroke patients scoring outside anxiety range; or changes from baseline on an anxiety rating scale

See comment

See comment

Not estimable

196
(3 studies)

⊕⊝⊝⊝
very lowc,d

Statistically significant difference in anxiety scores on HADS‐A scale at 3 months, with reduction in anxiety for those using therapeutic CD (P value = 0.001); statistically significant differences in HAM‐A scores at 6 weeks and 4 weeks with reduced anxiety for those taking paroxetine and paroxetine with psychological therapy and those taking buspirone, respectively (P value < 0.01)

Co‐morbid depression

See comment

See comment

Not estimable

175
(2 studies)

See comment

Reduction in depression symptoms according to HAM‐D at 6 weeks and at 4 weeks for those taking paroxetine and paroxetine with psychological therapy and those taking buspirone, respectively

Quality of life ‐ not reported

See comment

See comment

Not estimable

See comment

Outcome not reported in any study

Social activities ‐ not reported

See comment

See comment

Not estimable

See comment

Outcome not reported in any study

Activities of daily living

See comment

See comment

Not estimable

81
(1 study)

⊕⊝⊝⊝
very lowb,e

Improvement in activities of daily living in all groups, but greatest improvement in those taking paroxetine with psychological therapy

Principal caregiver burden ‐ not reported

See comment

See comment

Not estimable

See comment

Outcome not reported in any study

*The basis for the assumed risk (e.g. median control group risk across studies) is provided in footnotes. The corresponding risk (and its 95% confidence interval) is based on the assumed risk in the comparison group and the relative effect of the intervention (and its 95% CI)
CI: confidence interval; RR: risk ratio

GRADE Working Group grades of evidence
High quality: Further research is very unlikely to change our confidence in the estimate of effect
Moderate quality: Further research is likely to have an important impact on our confidence in the estimate of effect and may change the estimate
Low quality: Further research is very likely to have an important impact on our confidence in the estimate of effect and is likely to change the estimate
Very low quality: We are very uncertain about the estimate

aStudy was unblinded, with further risks of recruitment bias and drop‐outs. Downgraded one level for risk of bias
bOnly one study with small number of participants, downgraded two levels for imprecision
cLimited detail in Wang 2005 and Zhang 2005 for effective assessment of bias; lack of blinding, risks of recruitment bias, and drop‐outs in Golding 2016. Downgraded two levels for risk of bias
dOnly three studies with few participants, all with different interventions that are not comparable. Downgraded one level for indirectness and one level for imprecision
eLimited detail in studies, unable to effectively assess risk of bias; downgraded one level

Figures and Tables -
Summary of findings for the main comparison. Interventions for treating anxiety after stroke
Comparison 1. Pharmacological agents versus control

Outcome or subgroup title

No. of studies

No. of participants

Statistical method

Effect size

1 Dizziness Show forest plot

2

153

Risk Ratio (M‐H, Random, 95% CI)

7.32 [0.96, 55.95]

Figures and Tables -
Comparison 1. Pharmacological agents versus control