The effect of psychosocial interventions on depression, anxiety, and quality of life in hemodialysis patients: a systematic review and a meta-analysis

The objectives of the present review are:

We considered eligible: (i) all the randomized controlled trials (RCTs) or quasi-randomized controlled trials (quasi-experimental) conducted in any healthcare infrastructures; (ii) all the psychosocial interventions administered on adults, defined as any intervention used to modify behavior, emotional state or feeling. We defined psychosocial interventions according to criteria used in previous Cochrane reviews of such interventions for long-term diseases [13, 16, 17]. This definition included any intervention involving systematic treatment with some form of feedback or review, including psychotherapy, psychoeducation, and interventions containing elements, such as education, goal-setting, cognitive restructuring, behavioral strategies, relaxation training, stress management, or support groups. Multimodal interventions, consisting of both psychological and other therapeutic procedures (e.g., exercise), were also included, if one or more social–psychological therapeutic technique was clearly evident. Both individual and group-focused interventions were included, regardless of delivery mode; (iii) interventions involving people affected from CKD of any etiology and duration, stages 3–5 (eGFR falling below 60 ml/min/1.73m²), undergoing hemodialysis, at home or an ambulatory; (iii) studies that stated a validated self-reported measure of depression and/or anxiety; (iv) studies that compared an intervention group with usual care or placebo. Only studies available in the full text were included to ensure sufficient appraisal and review.

Studies were identified by searching the following databases: PsycInfo, Medline, Embase, Web of Science (WoS), Cumulative Index to Nursing and Allied Health Literature (CINAHL), and Cochrane Central Register of Controlled Trials (CENTRAL) and Scopus. In addition, the eligible articles’ references were screened to identify any relevant articles missed by the electronic and manual searches. We screened all the available articles in the databases until the 30th of June 2022.

The search strategy included a combination of MeSH terms and keywords with appropriate Boolean operators (Supplemental File 1). The quality of the search strategy, tailored for each database, was developed with the aid of a medical librarian with expertise in database searches for reviews. No time limits were applied across the searches, and all the articles were screened with the support of the reference's manager Zotero (v5.0). The search was restricted to articles published in English, Italian or Chinese.

Four authors (SB, MA, GA, EG) independently reviewed the articles by title, abstract and full text to identify all the eligible studies. Disagreements were solved by involving one or more additional authors (SB, MA, GA, EG). We included all the RCTs and quasi-experimental on adults with CKD that assessed the efficacy of a psychosocial intervention on anxiety or depression. We excluded: (i) all the uncontrolled intervention studies, or case series studies, or observational studies; (ii) all the studies on children or adolescents; (iii) all the studies on people who were enrolled for kidney transplantation or had just received it; (iv) primary stages of CKD when eGFR is still normal or slightly reduced (eGFR above 60 ml/min/1.73 m²); (v) all the pharmaceutical interventions, the exclusive exercise interventions, the spiritual interventions, alternative medicine interventions and interventions revolving around making a specific change in the diet; (vi) all the studies that reported a dichotomous measurement of anxiety or depression.

Four authors (SB, MA, GA, EG) independently extracted the data for the qualitative synthesis, the quantitative synthesis (SB, MA, GA, EG) and the Risk of Bias assessment (SB, MA, GA, EG). Data from the included studies were extracted using a pre-designed data extraction form developed from the data collection checklist by the Cochrane Effective Practice and Organisation of Care Review Group (EPOC). Key aspects of data extraction included: author, publication date, country, demographic or disease characteristics, study design, intervention description, the deliverer of the intervention, data analyses, and results for the outcomes of interest. Where missing data were encountered, corresponding authors were contacted to retrieve further information. We found no papers reporting on the same trial, so only unique evidence was gathered.

Validated psychometric self-report measures of depression and anxiety were extracted, such as the Beck Depression Inventory (BDI) scale or the Hospital Anxiety and Depression Scale (HADS). Whether there were missing or unclear data, we tried to contact the authors for further clarifications. To synthetize the key characteristic of the psychosocial interventions, two authors independently extracted the tools and strategies applied in each intervention. At the end of the synthesis process, all the authors reviewed and agreed on the developed categorization.

The risk of publication bias was evaluated to control whether studies reporting statistically significant results were more likely to be published, potentially overestimating the real effect size [20]. To assess the presence of publication bias, a funnel plot for symmetry was constructed, with asymmetry suggesting the possibility of publication bias. Given the subjective nature of interpreting funnel plots, Egger's test to measure funnel plot asymmetry was used as an objective test to support the conclusions [21].

A narrative synthesis of the included studies was conducted to provide a conceptual summary of the findings across all the included studies. A quantitative synthesis was conducted using R’s metan package (v.1.11.0). We summarized the findings using descriptive statistics, including mean, median, standard deviations (SDs) and frequencies (percents). A DerSomonian–Laird (DerSimonian & Laird) random-effects meta-analysis with 95% confidence intervals was performed. A random-effects model output was selected as moderate or high clinical and methodological heterogeneity was expected. Forest plots were provided for anxiety, depression and quality of life. Since the variables of interest were all continuous but reported using different weight units, a standardized mean difference (SMD) was computed.¹⁴ When standard deviation were not reported, we converted standard errors (SEs), 95% confidence intervals (Cis) and interquartile ranges (IQRs) to SDs using the formulas suggested in the Cochrane Handbook for Systematic Reviews and Meta-Analysis. The Hedges’G (g) form of the SMD was used. For anxiety and depression, negative SMDs indicate a difference in results favoring the intervention. A SMD of 0.20 is considered a small, 0.50 medium and > 0.80 a large effect [22]. Depression and anxiety were recoded where necessary, so low scores indicate lesser depression or anxiety. Cochrane's Q chi-squared test and I² statistic were used to assess the level of statistical heterogeneity. A p value of < 0.1 and a I² > 75% were considered significant. Sensitivity analysis were performed after the detection and the examination of outliers, using R’s InfluenceAnalysis function [23]. Subgroup analyses were also conducted to assess if effect sizes vary according to the following characteristics: (i) type of intervention, (ii) deliverer, (iii) length of the intervention (more or less than 5 weeks), (iv) type of sessions (group vs. individual). Sensitivity and subgroup analyses were conducted on the most complete data set, using endpoint-point data of each study. If there were fewer than three studies for each subgroup, we decided not to pool the subgroup [24].

The overall certainty of evidence across studies was rated using the Grading of Recommendations Assessment, Development and Evaluation (GRADE) working group guidelines. According to the corresponding evaluation criteria, the quality of evidence was classified into four categories: high, moderate, low and very low.

The electronic searches resulted in 7320 citations. After the merging of the duplicates, 5324 citations were screened. Four authors (SB; GA; EG; CB) screened the titles and abstracts and identified 193 records as potentially relevant, which were retrieved for further assessment. After full-text screening, 15 studies were deemed eligible according to the inclusion and exclusion criteria shown in Fig. 1. The most common reasons for exclusion after full-text appraisal were: (i) not measuring the incidence of depression or anxiety as an outcome [25‐31], (ii) not testing a psychosocial or cognitive intervention [32‐35], and (iii) not a RCTs or quasi-experimental controlled study design [36‐38]. Out of the 15 trials included in our systematic review and meta-analysis, 10 were RCTs, and five were quasi-experimental. All the RCTs randomized the participants at individual levels. A summary of the study characteristics and demographics is presented in Table 1 [39‐54].

Fifteen trials were included, with a total of 837 participants [39‐54]. Trial sample sizes ranged from 22 to 94 participants. The trials were conducted in six different countries (Table 1), with the Middle East (e.g., Iran) and Asian (e.g., Taiwan) countries being the most prevalent. The trials were published between 2003 and 2020, with 60% of the included studies published in the last 5 years. With regard to demographic characteristics of the samples, females (58%) were slightly more prevalent than males across the studies. The samples’ mean age ranged from 41.75 years [47] to 63.01 years [51], with a weighted mean age of 51.5 years. Four studies did not report the mean age for the participants, and three studies provided little information on the sample characteristics (Table 1). All the included studies involved dialysis patients, with a mean length of dialysis ranging from 3 months to more than 5 years. Few studies reported the main cause of the disease, [41, 46, 51] with diabetes and high blood pressure being the most common. Both depression and anxiety were a primary outcome in all the included studies. Most of the studies used or the Beck Inventory Scale (BDI) or Hospital Anxiety and Depression scale (HADS) to measure depression, while only three studies used the Depression Anxiety Stress Scales (DASS). Instead, anxiety was prevalently measured with either HADS or DASS. Only four studies assessed outcomes on multiple timing after the intervention, while the remaining studies used a single assessment (Table 1). Twelve studies assessed outcomes shortly after the intervention (less than 6 weeks), while the other assessed outcomes at longer intervals following the intervention for a maximum period of 3 months (Table 1). Only two studies stated a follow-up, ranging from 2 up to 9 months [41, 47]. Out of the 15 studies, 13 found significantly lower depression scores in the intervention groups compared to the control groups.

Great variability existed in relation to intervention contents and structure, as shown in Table 2. Frequency and duration of treatment sessions varied considerably among the trials, with a minimum number of four sessions up to 12 sessions (Table 2). The length of the session varied from four to 12 weeks (Table 2). All the interventions were delivered face-to-face, either individually and/or in a group format (Table 2). Across the interventions, the sessions lasted between 15 and 120 min. The interventions were usually delivered immediately before or the day after a dialysis treatment. The majority of the sessions were held in dialysis centers or at the participant's home (Table 2). Many studies used tools (e.g., booklets) or devices (e.g., laptops) to enhance the interventions' effect. The most common provider for the interventions were nurses, followed by multidisciplinary teams, and finally psychologists (Table 2). Only one study [45] used a peer (i.e., an expert patient) as the main provider for the intervention. When the interventions were coded (Table 2), the following categories were defined: psychoeducation, psychological support and relaxation-based therapy. The psychoeducational intervention aimed to increase the participant's health literacy and develop goal-setting strategies and/or problem-solving skills (Table 2). The interventions coded as psychological support interventions were generally aimed to promote the patients' acceptance and elaboration of the disease; they were the majority and mainly cognitive–behavioral interventions (CBTs), with only one study testing a Schneider's hope therapy intervention [50]. The relaxation-based interventions aimed to increase self-regulation and emotional, cognitive, and behavioral flexibility were the most heterogeneous category, enclosing different techniques, such as Benson relaxation-based methods or guided imagery intervention or breathing training (Table 2).

In relation to the comparison group, most studies compared the intervention to usual care or handbook provision (Table 2), with only one study using an enhanced comparison group [50].

To assess the quality of evidence for outcomes, the Grade framework was performed and determined the effect of psychological intervention to be of moderate quality on quality of life (Table 3). The evidence about depression and anxiety was downgraded to low (Table 3).