Introduction: People living with chronic diseases are at an increased risk of anxiety and depression, which are associated with poorer medical and psychosocial outcomes. Many studies have examined the trajectories of depression and anxiety in people with specific diseases, including the predictors of these trajectories. This is valuable for understanding the process of adjustment to diseases and informing treatment planning. However, no review has yet synthesised this information across chronic diseases. Methods: Electronic databases were searched for studies reporting trajectories of depression or anxiety in chronic disease samples. Data extracted included sample characteristics, results from trajectory analyses, and predictors of trajectories. Meta-analysis of the overall pooled prevalence of depression and anxiety trajectories was conducted, and qualitative synthesis of disease severity predictors was undertaken. Results: Following search and screening, 67 studies were included (N = 61,201 participants). Most participants followed a stable nonclinical trajectory for depression (69.0% [95% CI: 65.6, 72.2]) and anxiety (73.4% [95% CI: 66.3, 79.5]). Smaller but meaningful subsamples followed a trajectory of depression and anxiety symptoms consistently in the clinical range (11.8% [95% CI: 9.2, 14.8] and 13.7% [95% CI: 9.3, 19.7], respectively). Several clinical and methodological moderators emerged, and qualitative synthesis suggested that few aspects of disease severity were associated with participants’ trajectories. Conclusion: Most people with chronic disease follow a trajectory of distress that is low and stable, suggesting that most people psychologically adjust to living with chronic disease. Evidence also suggests that the nature and severity of the disease are not meaningful predictors of psychological distress.

Approximately half the adult population lives with one or more chronic diseases [1], which require them to cope with a range of day-to-day challenges, such as difficult symptoms, loss of independence, and daily self-management. Because of these challenges, people with chronic disease are at significantly increased risk of depression and anxiety, compared to the general population [2‒4]. In addition, the co-occurrence of psychological distress in chronic disease is associated with a range of adverse outcomes, such as poorer self-rated health [5], increased work absenteeism [6], and poorer disease self-management and prognosis, including lower adherence to treatments and lifestyle recommendations [7, 8]. As such, there is great interest in promoting successful psychological adjustment after the onset of chronic disease, which is commonly described as the absence of excessive psychological distress [9‒11].

Fortunately, evidence suggests that psychotherapy approaches successfully reduce distress in the context of chronic disease [12], thus promoting successful adjustment. However, these treatments only appear to be efficacious among people with clinically elevated symptoms of depression or anxiety, as opposed to unselected samples [12]. Importantly, many individuals appear to experience heightened symptoms of depression or anxiety after the onset of a chronic disease that remits on its own (e.g., [13, 14]). This pattern of distress highlights the concept of psychological adjustment as a return to equilibrium after a stressor, whereby people utilise their resources to achieve a “new normal” [11, 15]. In such cases, offering psychological treatment may not be clinically required or cost-effective. However, it is not clear what proportion of people with chronic disease experience consistently elevated distress and when the need for treatment may emerge in the disease trajectory. A clear understanding of the proportion and timing of patients’ psychological adjustment is important to appropriately implement and successfully target psychological interventions within our health systems.

To understand patterns of psychological adjustment in the disease trajectory, researchers have conducted longitudinal studies that measure peoples’ symptoms over time. Recent advances in analytic techniques have allowed researchers to better capture the heterogeneity of individuals’ adjustment by using analytic methods such as group-based trajectory modelling and growth-mixture modelling [16, 17]. These methods allow researchers to identify and model distinct groups that follow similar symptom changes (i.e., their trajectories) longitudinally [16, 18]. A key advantage of these approaches is that they capture the variability of individuals’ adjustment to chronic disease better over time, compared to studies that simply report high-level aggregate group change following disease onset [16].

These statistical techniques have been successfully adopted in people with chronic disease, such as breast cancer [19], cardiovascular disease [20], and spinal cord injury [21]. Interestingly, these studies show similar patterns of adjustment trajectories. Specifically, most patients in these samples show a trajectory of symptoms that is low and stable (often classed as a “resilient” trajectory). Additionally, many of these studies show a subgroup of participants following a trajectory of consistently high distress (i.e., often classed as “chronic,” and evidence of poor adjustment to illness). Many studies examining the trajectories of distress within chronic diseases also examine whether certain demographic or illness-related factors predict participants’ adjustment trajectories, such as age, gender, or disease severity (e.g., [22‒24]). These predictor analyses can provide useful insights into who may be at greater risk of poor adjustment and why.

To the best of our knowledge, no study has synthesised this information across different health conditions, despite the significant practical and theoretical value it may carry. It is valuable to understand who may require psychological support after the onset of chronic disease and when such a need may emerge. In addition, understanding known risk factors may help clinicians identify which patients may be more, or less, likely to become distressed following disease onset. It is also important to understand whether different disease types or characteristics predict certain adjustment trajectories. For example, it may be that people with a disease characterised by progressive decline are more likely to follow a worsening distress trajectory, compared to those with a mild, constant, or remitting trajectory. However, it is difficult to answer such questions within individual studies. Meta-analytic methods overcome many of these limitations, as they allow for the systematic investigation of between-study moderators, such as chronic disease types [25, 26].

This systematic review and meta-analysis sought to synthesise the available literature and examine trajectories of depression and anxiety in chronic disease populations. Our primary aim was to understand the number of trajectories, the shape of the trajectories, and the proportion of people falling into different trajectories. Second, we aimed to understand what factors predicting adjustment trajectories have been investigated.

This systematic review and meta-analysis was prospectively registered on the Open Science Framework repository (https://osf.io/hfy6k) and conducted according to PRISMA guidelines [27].

Selection Criteria

Included articles were those conducted in participants with any chronic disease (e.g., cancer, chronic pain, and mixed populations) and who measured participants’ symptoms of depression or anxiety using a valid measure. Studies were included if they used a longitudinal design with at least three time-points and if they used analytic methods that identified distinct trajectories of change over time (e.g., growth mixture modelling, latent class growth analysis).

Studies examining the trajectories of depression and anxiety after acute injury or illness were excluded, unless the injury was likely to result in ongoing health and psychosocial complications (e.g., spinal cord injury, burn injury). Additionally, studies that measured trajectories of depression and anxiety within the context of psychiatric treatment were excluded (e.g., where all participants were enrolled in psychotherapy or receiving an antidepressant medication), given they may alter the natural course of adjustment and distress. Other psychosocial constructs, general stress, coping, and social support were outside the scope of this review. Finally, studies were eligible if they were original articles published in English. Both peer-reviewed journal articles and dissertations were included, while secondary or subgroup analyses of previously reported data were excluded.

Search Strategy and Study Selection

A systematic search was conducted in the electronic databases PsycINFO, Medline, and Embase. Searches were conducted from inception up to January 2023. Search terms included terminology related to (1) depression and anxiety and (2) trajectories. See supplementary Table 1 (for all online suppl. material, see https://doi.org/10.1159/000533263) for the search terms used. The search strategy was chosen to maximise sensitivity; therefore, stems related to chronic disease were not included. We also examined the reference lists of included studies (i.e., backward search) and searched for more recent articles citing the included studies (i.e., forward search) to identify other potentially eligible papers. Finally, any relevant systematic review or meta-analysis studies were identified, and their included studies were examined for potential relevance.

Following the systematic search, titles and abstracts of all identified records were reviewed by two researchers, with a randomly selected 10% of records double-screened in order to ensure agreement and consistency with inclusion and exclusion criteria. Following this, two authors reviewed all full-text articles to determine final eligibility. The strength of inter-rater agreement was determined as moderate if Kappa (κ) was between 0.41 and 0.60, as substantial if between 0.61 and 0.80, and as nearly perfect if >0.81 [28]. Agreement for title and abstract screening was substantial (κ = 0.79) and full-text review was nearly perfect (κ = 0.91). Any disagreements at the full-text stage were resolved via consensus.

Data Extraction

Data were extracted by two authors into a predefined spreadsheet that had been piloted with 10 relevant articles. A randomly selected 10% of data were extracted by both authors to ensure consistency and accuracy. The information extracted included the study characteristics (e.g., year of publication, country, recruitment setting), sample characteristics (e.g., N, health condition, other population specifiers, whether the sample was derived from a population survey), methodology (e.g., relevant inclusion/exclusion criteria, study duration, number of assessments), and analyses (e.g., analytic approach, missing data method, attrition).

Outcomes

The primary outcomes of interest were trajectories of depression or anxiety symptoms. Additional relevant information extracted included the outcome measure used, total number of trajectories, number of stable trajectories, and number of transient trajectories. Additionally, the original labels given to each trajectory and the proportion of participants in each trajectory were extracted. In some instances, measures were introduced to ensure consistency in the reporting and comparability of the results regarding the shape of trajectories. For example, some authors described a trajectory shape as transient, despite no meaningful change being observed (e.g., describing an average 1-point increase on a 27-point scale as “increasing”). Therefore, for the purpose of this study, fluctuating trajectories were relabelled as stable where no meaningful change was observed, defined as >30% of the baseline score.

Predictors

The predictors of trajectories examined by each study were also extracted and included demographic (e.g., age), psychological (e.g., proportion receiving psychotropic medication or psychotherapy, history of psychiatric disorders), medical (e.g., frequency of hospitalisations in the last year), disease severity, behavioural/self-management (e.g., smoking status), and other predictors. As a result of the huge diversity in the examined trajectories and analyses used, we chose to focus on any predictors that were considered indicators of disease severity (e.g., cancer stage).

Study Quality

Study quality and risk of bias were examined using a modified version of the Newcastle-Ottawa Scale (NOS) for cohort studies. The items used were those relating to participant selection, comparability, and outcome. Regarding participant selection and comparability, we examined whether the participants were recruited from community, primary, secondary, or tertiary settings. We also coded whether authors reported the proportion of participants consenting to the research of those who were approached. Regarding outcome measurement, we examined whether authors reported attrition rates and how missing data were examined or handled in the analyses.

Data Analysis

To analyse and compare results between studies, the trajectories from individual studies were combined as follows. First, the combined prevalence of participants belonging to trajectories that remained below the instrument’s clinical cut-off was calculated, forming a “nonclinical” trajectory for each paper. Second, the combined prevalence of participants belonging to trajectories that remained above the clinical cut-off was calculated, forming a combined “clinical” trajectory. Comprehensive Meta-Analysis (CMA V3) was then used to calculate the pooled mean prevalence of participants belonging to each of the above two trajectories for depression and anxiety. Random effects models were used based on the assumption that there would be true heterogeneity between studies, given that studies were conducted in different populations and settings.

Heterogeneity and uncertainty in outcomes were examined in multiple ways. Firstly, Cochran’s Q and its associated p value were reported, though it is noted that this test can be highly sensitive in meta-analyses with larger samples of studies [29]. In addition, the I2 statistic was reported for outcomes. This statistic reports the percentage of variance in true effects rather than due to sampling error [30]. Because meta-analyses of prevalence often yield high I2 statistics that may not indicate true heterogeneity [31], we also reported prediction intervals, which present the expected range of outcomes in 95% of similar studies and indicate the degree of uncertainty in the obtained estimate [32].

Subgroup analyses were also conducted to explore possible sources of heterogeneity and to understand whether participants’ trajectories differed by a number of potential moderators: chronic disease type, recruitment setting (e.g., primary, secondary, tertiary), outcome measure used, method of analysis (growth mixture modelling vs. latent class growth analysis/group-based trajectory modelling), and study country. Mixed effects models were used for subgroup analysis, which pools studies within subgroups using random effects but tests for differences between subgroups using fixed effects. Subgroup analysis was only performed when there were adequate data for each moderator. This was defined as a minimum of 3 studies and 100 participants per moderator. In addition, meta-regression analysis was used to understand whether sample size and study year were associated with study outcomes.

Finally, sensitivity analyses were conducted removing outliers, which we defined as any study whose prevalence estimate fell outside +/− 2 standard deviations of the overall pooled prevalence for each outcome. Both the main and moderator analyses were re-run with outliers removed.

Study Selection

After conducting the systematic search, retrieving records, and removing duplicates, 7,764 titles and abstracts were reviewed (of which 7,474 were excluded). After full-text review with the remaining 290 studies, 67 remained eligible for inclusion and data extraction (see Fig. 1).

Fig. 1.

PRISMA diagram.

Study Characteristics

See Table S2 for a summary of included studies. A total of 67 studies were included, with sample size ranging from 74 to 4,803 (total N = 61,201). Overall, 31 health conditions were investigated in participants from 13 countries, mainly the USA (k = 28), Australia (k = 10), the Netherlands (k = 8), and China (k = 4). The most commonly studied health conditions were cancer (k = 11), cardiovascular disease (k = 12), chronic pain (k = 6), diabetes (k = 5), human immunodeficiency virus (HIV; k = 4), liver disease (k = 3), spinal cord injury (SCI; k = 4), and traumatic brain injury (TBI; k = 3). Of the included studies, 37 drew participants from existing cohorts, such as the Health and Retirement Study [20], the South London Stroke Register [33], and the Montreal Diabetes Health and Well-Being Study [34]. Meanwhile, 17 papers included original samples. Most studies (k = 31) recruited participants via a tertiary setting (e.g., hospitals); 19 studies recruited participants via community or population-based settings, 12 via secondary settings (e.g., specialist), and 1 via primary setting (e.g., general practice). Most (k = 57) of the included studies reported depression trajectories, while 27 reported anxiety trajectories (17 studies reported on both depression and anxiety). Most studies (k = 41, 61%) reported trajectories over a duration of 2 years or more, with the maximum duration reaching 18 years [35]. In comparison, less studies measured trajectories over 1 year or less (k = 23, 34%). The smallest duration of measurement was 5 months [36]. Three studies did not report the measurement duration, or the duration was unclear. Overall, the average duration of trajectory measurement across studies was 3.76 years (approximately 3 years and 9 months). In terms of measurement frequency, four studies (6%) collected data every month, while 24 studies (36%) collected data annually. Other studies were more variable, beginning with monthly or bi-monthly assessments before increasing the time between measurements to once a year (k = 36, 54%). Overall, the measurement frequency ranged from 3 weeks [21] to 5.5 years [37]. Some studies (k = 20, 30%) assessed whether participants were in receipt of mental health care during the study period. Among these studies, approximately 4–13% of participants were undergoing psychotherapy, while approximately 3–48% of participants were taking psychotropic medication.

Study Quality and Methods

See Table S3 for a summary of study methods and statistical approach. In the studies assessing depression (k = 57), most recruited participants were from tertiary settings (k = 20), followed by community settings (k = 12), secondary settings (k = 9), population-based settings (k = 2), and primary settings (k = 1). One study used a mixed-recruitment method and recruited participants from primary, secondary, and tertiary settings. Eighteen studies reported the proportion of participants consenting to the research of those who were approached; the proportion consenting ranged from 55% to 96%. Furthermore, 40 studies reported attrition rates and 19 examined or managed missing data (e.g., via multiple imputation methods).

In studies assessing anxiety (k = 27), most recruited participants were from tertiary settings (k = 10), followed by secondary settings (k = 6), community settings (k = 2), and primary settings (k = 1). One study used a mixed-recruitment method and recruited participants from community and tertiary settings. Eleven studies reported the proportion of participants consenting to the research of those who were approached; the proportion consenting ranged from 55% to 96%. Furthermore, 16 studies reported attrition rates and seven examined or managed missing data.

Depression

There were 57 included studies reporting depression trajectories, representing 49,926 participants. Of these, three studies treated their male and female participants as separate samples when performing trajectory analyses. Therefore, these samples will be treated as separate studies for the purpose of reporting the depression trajectory results, increasing the total number of depression samples to 60. Of these 60 samples, most (k = 24) reported a 3-trajectory solution, while 19 reported a 4-trajectory solution. The remaining studies reported on a 2-trajectory solution (k = 7), 5-trajectory solution (k = 5), 6-trajectory solution (k = 2), and 7-trajectory solution (k = 1) (see Table S4).

Several common trajectories were found across studies. For example, most studies (k = 55, 92%) identified one or more trajectories that remained in the nonclinical range, with 4–89% of participants belonging to these trajectories (see Fig. 2). Most studies (k = 45, 75%) also identified one or more trajectories that remained in the clinical range, with 1–43% of participants belonging to these trajectories. Furthermore, some studies (k = 24, 40%) identified stable moderate trajectories, in which participant symptom scores were close to the measure’s cut-off. The proportion of participants in these trajectories ranged from 10% to 56%.

Fig. 2.

Depiction of depression trajectories.

Fig. 2.

Depiction of depression trajectories.

Close modal

The number and shape of transient trajectories were more heterogeneous. Some studies (k = 28, 47%) reported increasing trajectories. Of these increasing trajectories, most (k = 17, 61%) increased from the nonclinical to clinical range (see Table S4). Furthermore, approximately half (k = 31, 52%) of the studies reported decreasing symptom trajectories. Among these decreasing trajectories, most (k = 20, 65%) decreased from the clinical to nonclinical range. Finally, some studies (k = 17, 28%) reported trajectories that followed a fluctuating pattern (e.g., increasing and then decreasing again, or vice versa).

Meta-Analysis

The overall pooled prevalence of participants following a nonclinical depression trajectory was 69.0% (95% CI: 65.6, 72.2), with a high amount of heterogeneity observed (Q56 = 3,087.5, p < 0.001, I2 = 98.2, τ2 = 0.3). The moderating effect of health condition was significant (Q7 = 34.8, p < 0.001), as was the study country (Q3 = 8.44, p = 0.04) (see Table 1). The moderating effect of depression outcome measure was nonsignificant (Q4 = 1.36, p = 0.85), as was the effect of analysis type (Q1 = 2.43, p = 0.12), recruitment setting (Q2 = 5.14, p = 0.08), sample size (B = 0.0001, 95% CI: −0.0001, 0.0003, Q1 = 1.68, p = 0.24), and study year (B = −0.03, 95% CI: −0.06, 0.01, Q1 = 1.82, p = 0.18).

Table 1.

Meta-analysis results

NonclinicalClinical
NcPooled prevalence (95% CI)I2T2PIp valueNcPooled prevalence (95% CI)I2T2PIp value
Depression 
 Overall 57 69.0 (65.6, 72.2) 98.2 0.3 41.8, 87.3  57 11.0 (8.7, 13.9) 98.6 0.9 1.9, 45.8  
Type of chronic condition    <0.001      <0.001 
 Cancer 11 67.9 (57.1, 77.0) 98.9 0.6 25.3, 93.0  10 6.0 (3.2, 11.1) 98.0 0.8 0.7, 36.6  
 Cardiovascular 12 79.4 (72.6, 84.9) 98.0 0.4 46.9, 94.4  11 11.2 (7.2, 17.1) 97.1 0.5 0.0, 0.5  
 Chronic pain 68.6 (56.6, 78.6) 98.5 0.4 24.6, 93.6  12.1 (8.4, 17.0) 96.2 0.2 3.4, 35.1  
 Diabetes 79.7 (70.5, 86.5) 97.5 0.3 36.5, 96.4  7.7 (4.6, 12.4) 92.5 0.3 0.6, 54.0  
 HIV 61.1 (56.0, 66.0) 94.8 0.0 49.7, 71.4  10.9 (4.0, 26.6) 99.5 1.2 0.1, 96.0  
 Liver disease 49.5 (22.5, 76.9) 97.6 1.1 0.0, 100.0  26.5 (18.5, 36.5) 82.2 0.1 0.0, 99.0  
 SCI 59.9 (52.2, 67.1) 75.6 0.1 24.5, 87.3  11.1 (3.8, 28.3) 94.9 1.1 0.1, 95.7  
 TBI 72.3 (68.6, 75.8) 56.4 0.0 45.2, 89.2  13.8 (6.4, 27.1) 96.0 0.5 0.0, 100.0  
Recruitment setting   0.08      0.15 
 Community 13 73.9 (66.8, 79.9) 98.8 0.0 65.9, 80.6  13 10.6 (6.1, 17.7) 99.3 1.1 1.0, 57.7  
 Secondary 62.4 (53.2, 70.8) 98.6 0.0 51.3, 72.4  15.2 (9.8, 22.8) 98.1 0.5 2.8, 53.0  
 Tertiary 25 72.6 (67.7, 77.0) 96.1 0.0 67.4, 77.2  24 9.4 (6.1, 14.4) 97.8 1.2 1.0, 51.5  
Analysis      0.32      0.48 
 LCGA 19 68.2 (67.5, 69.0) 97.8 0.3 39.9, 87.4  19 15.1 (14.5, 15.8) 98.3 0.76 2.7, 52.8  
 GBTM 17 68.4 (67.7, 69.1) 98.6 0.3 39.8, 87.6  17 14.3 (13.8, 14.9) 98.9 0.89 2.2, 55.0  
 GMM 19 72.2 (66.3, 77.5) 98.3 0.3 28.5, 94.4  19 8.0 (5.2, 12.0) 98.3 0.8 3.5, 17.3  
Outcome measure     0.85      <0.001 
 BDI 70.8 (61.7, 78.5) 95.2 0.3 20.7, 95.7  20.0 (14.6, 26.6) 91.8 0.2 6.1, 48.9  
 CES-D 22 65.7 (59.5, 71.4) 98.7 0.4 24.3, 91.9  22 14.2 (9.7, 20.4) 99.1 1.0 5.5, 32.1  
 HADS-D 14 68.5 (60.2, 75.8) 97.5 0.4 23.8, 93.8  14 4.7 (2.4, 8.8) 97.7 1.2 1.8, 11.7  
 PHQ-9 68.9 (61.4, 75.5) 97.9 0.2 21.5, 94.7  7.0 (5.7, 8.6) 84.7 0.1 1.8, 11.7  
Study country     0.04      0.001 
 Australia 73.1 (68.1, 77.6) 95.5 0.1 54.2, 86.2  8.3 (6.6, 10.4) 83.4 0.1 3.8, 17.2  
 China 61.2 (54.5, 67.6) 93.7 0.1 26.2, 87.5  24.2 (13.4, 39.7) 98.5 0.4 1.3, 88.2  
 Netherlands 68.6 (53.9, 80.3) 99.1 0.6 17.6, 95.7  8.3 (4.3, 15.5) 97.7 0.7 0.7, 53.2  
 USA 25 69.4 (62.7, 75.3) 98.6 0.5 33.7, 91.0  25 13.9 (10.1, 18.9) 98.2 0.7 2.7, 48.7  
ANXIETY             
 Overall 26 73.4 (66.3, 79.5) 98.3 0.7 29.7, 94.7  26 13.7 (9.3, 16.7) 97.8 1.0 1.9, 56.8  
Type of chronic condition   0.35      0.8 
 Cancer 69.5 (51.9, 82.8) 98.4 1.0 14.2, 96.9  13.5 (6.8, 25.1) 98.1 1.0 1.1, 68.3  
 Cardiovascular 78.8 (64.7, 88.2) 98.3 0.9 21.5, 98.1  11.6 (4.1, 28.6) 98.5 1.9 0.3, 85.9  
Recruitment setting   0.11      <0.001 
 Community 85.5 (44.4, 97.7) 99.2 3.1 0.0, 100.0  2.2 (0.2, 19.2) 97.7 3.8 0.0–100  
 Secondary 66.9 (58.3, 74.5) 93.8 0.2 34.5, 88.6  25.7 (17.6, 35.7) 94.4 0.3 6.1, 64.7  
 Tertiary 14 78.7 (69.5, 85.7) 98.3 0.8 32.9, 96.5  14 6.9 (3.3, 13.9) 97.7 1.8 0.4, 61.0  
Analysis      0.70      0.01 
 LCGA 68.9 (56.2, 79.3) 98.5 0.68 22.0, 94.6  21.0 (10.9, 35.8) 98.3 1.22 1.7, 80.8  
 GBTM 74.1 (58.0, 85.6) 95.1 0.66 14.3, 98.0  23.2 (11.7, 40.7) 93.7 0.71 1.5, 85.8  
 GMM 12 75.5 (62.9, 84.8) 98.1 0.8 33.1, 97.5  12 6.3 (3.0, 12.6) 98.1 1.5 2.0, 17.9  
Outcome measure     0.07      0.09 
 GAD-7 83.9 (75.4, 89.8) 96.4 0.3 7.0, 99.7  0.3 (0.0, 18.7) 94.9 18.9 0.0, 98.5  
 HADS-A 11 72.2 (60.0, 81.8) 98.6 0.8 23.4, 95.7  11 12.0 (6.1, 22.4) 98.3 1.2 3.3, 35.1  
 STAI 63.7 (40.6, 81.8) 97.9 1.5 11.2, 96.1  22.5 (10.1, 42.9) 98.5 1.6 3.5, 70.2  
Study country     0.02      0.01 
 Australia 81.0 (70.8, 88.2) 97.0 0.5 37.5, 96.8  12.9 (8.2, 19.5) 86.7 0.3 3.1, 41.1  
 Netherlands 64.3 (49.2, 77.0) 97.8 0.6 15.0, 94.8  35.4 (19.6, 55.1) 98.5 1.0 0.7, 89.9  
 USA 49.4 (26.7, 72.3) 96.9 0.8 1.2, 98.8  30.3 (13.7, 54.5) 96.1 0.9 0.4, 97.8  
NonclinicalClinical
NcPooled prevalence (95% CI)I2T2PIp valueNcPooled prevalence (95% CI)I2T2PIp value
Depression 
 Overall 57 69.0 (65.6, 72.2) 98.2 0.3 41.8, 87.3  57 11.0 (8.7, 13.9) 98.6 0.9 1.9, 45.8  
Type of chronic condition    <0.001      <0.001 
 Cancer 11 67.9 (57.1, 77.0) 98.9 0.6 25.3, 93.0  10 6.0 (3.2, 11.1) 98.0 0.8 0.7, 36.6  
 Cardiovascular 12 79.4 (72.6, 84.9) 98.0 0.4 46.9, 94.4  11 11.2 (7.2, 17.1) 97.1 0.5 0.0, 0.5  
 Chronic pain 68.6 (56.6, 78.6) 98.5 0.4 24.6, 93.6  12.1 (8.4, 17.0) 96.2 0.2 3.4, 35.1  
 Diabetes 79.7 (70.5, 86.5) 97.5 0.3 36.5, 96.4  7.7 (4.6, 12.4) 92.5 0.3 0.6, 54.0  
 HIV 61.1 (56.0, 66.0) 94.8 0.0 49.7, 71.4  10.9 (4.0, 26.6) 99.5 1.2 0.1, 96.0  
 Liver disease 49.5 (22.5, 76.9) 97.6 1.1 0.0, 100.0  26.5 (18.5, 36.5) 82.2 0.1 0.0, 99.0  
 SCI 59.9 (52.2, 67.1) 75.6 0.1 24.5, 87.3  11.1 (3.8, 28.3) 94.9 1.1 0.1, 95.7  
 TBI 72.3 (68.6, 75.8) 56.4 0.0 45.2, 89.2  13.8 (6.4, 27.1) 96.0 0.5 0.0, 100.0  
Recruitment setting   0.08      0.15 
 Community 13 73.9 (66.8, 79.9) 98.8 0.0 65.9, 80.6  13 10.6 (6.1, 17.7) 99.3 1.1 1.0, 57.7  
 Secondary 62.4 (53.2, 70.8) 98.6 0.0 51.3, 72.4  15.2 (9.8, 22.8) 98.1 0.5 2.8, 53.0  
 Tertiary 25 72.6 (67.7, 77.0) 96.1 0.0 67.4, 77.2  24 9.4 (6.1, 14.4) 97.8 1.2 1.0, 51.5  
Analysis      0.32      0.48 
 LCGA 19 68.2 (67.5, 69.0) 97.8 0.3 39.9, 87.4  19 15.1 (14.5, 15.8) 98.3 0.76 2.7, 52.8  
 GBTM 17 68.4 (67.7, 69.1) 98.6 0.3 39.8, 87.6  17 14.3 (13.8, 14.9) 98.9 0.89 2.2, 55.0  
 GMM 19 72.2 (66.3, 77.5) 98.3 0.3 28.5, 94.4  19 8.0 (5.2, 12.0) 98.3 0.8 3.5, 17.3  
Outcome measure     0.85      <0.001 
 BDI 70.8 (61.7, 78.5) 95.2 0.3 20.7, 95.7  20.0 (14.6, 26.6) 91.8 0.2 6.1, 48.9  
 CES-D 22 65.7 (59.5, 71.4) 98.7 0.4 24.3, 91.9  22 14.2 (9.7, 20.4) 99.1 1.0 5.5, 32.1  
 HADS-D 14 68.5 (60.2, 75.8) 97.5 0.4 23.8, 93.8  14 4.7 (2.4, 8.8) 97.7 1.2 1.8, 11.7  
 PHQ-9 68.9 (61.4, 75.5) 97.9 0.2 21.5, 94.7  7.0 (5.7, 8.6) 84.7 0.1 1.8, 11.7  
Study country     0.04      0.001 
 Australia 73.1 (68.1, 77.6) 95.5 0.1 54.2, 86.2  8.3 (6.6, 10.4) 83.4 0.1 3.8, 17.2  
 China 61.2 (54.5, 67.6) 93.7 0.1 26.2, 87.5  24.2 (13.4, 39.7) 98.5 0.4 1.3, 88.2  
 Netherlands 68.6 (53.9, 80.3) 99.1 0.6 17.6, 95.7  8.3 (4.3, 15.5) 97.7 0.7 0.7, 53.2  
 USA 25 69.4 (62.7, 75.3) 98.6 0.5 33.7, 91.0  25 13.9 (10.1, 18.9) 98.2 0.7 2.7, 48.7  
ANXIETY             
 Overall 26 73.4 (66.3, 79.5) 98.3 0.7 29.7, 94.7  26 13.7 (9.3, 16.7) 97.8 1.0 1.9, 56.8  
Type of chronic condition   0.35      0.8 
 Cancer 69.5 (51.9, 82.8) 98.4 1.0 14.2, 96.9  13.5 (6.8, 25.1) 98.1 1.0 1.1, 68.3  
 Cardiovascular 78.8 (64.7, 88.2) 98.3 0.9 21.5, 98.1  11.6 (4.1, 28.6) 98.5 1.9 0.3, 85.9  
Recruitment setting   0.11      <0.001 
 Community 85.5 (44.4, 97.7) 99.2 3.1 0.0, 100.0  2.2 (0.2, 19.2) 97.7 3.8 0.0–100  
 Secondary 66.9 (58.3, 74.5) 93.8 0.2 34.5, 88.6  25.7 (17.6, 35.7) 94.4 0.3 6.1, 64.7  
 Tertiary 14 78.7 (69.5, 85.7) 98.3 0.8 32.9, 96.5  14 6.9 (3.3, 13.9) 97.7 1.8 0.4, 61.0  
Analysis      0.70      0.01 
 LCGA 68.9 (56.2, 79.3) 98.5 0.68 22.0, 94.6  21.0 (10.9, 35.8) 98.3 1.22 1.7, 80.8  
 GBTM 74.1 (58.0, 85.6) 95.1 0.66 14.3, 98.0  23.2 (11.7, 40.7) 93.7 0.71 1.5, 85.8  
 GMM 12 75.5 (62.9, 84.8) 98.1 0.8 33.1, 97.5  12 6.3 (3.0, 12.6) 98.1 1.5 2.0, 17.9  
Outcome measure     0.07      0.09 
 GAD-7 83.9 (75.4, 89.8) 96.4 0.3 7.0, 99.7  0.3 (0.0, 18.7) 94.9 18.9 0.0, 98.5  
 HADS-A 11 72.2 (60.0, 81.8) 98.6 0.8 23.4, 95.7  11 12.0 (6.1, 22.4) 98.3 1.2 3.3, 35.1  
 STAI 63.7 (40.6, 81.8) 97.9 1.5 11.2, 96.1  22.5 (10.1, 42.9) 98.5 1.6 3.5, 70.2  
Study country     0.02      0.01 
 Australia 81.0 (70.8, 88.2) 97.0 0.5 37.5, 96.8  12.9 (8.2, 19.5) 86.7 0.3 3.1, 41.1  
 Netherlands 64.3 (49.2, 77.0) 97.8 0.6 15.0, 94.8  35.4 (19.6, 55.1) 98.5 1.0 0.7, 89.9  
 USA 49.4 (26.7, 72.3) 96.9 0.8 1.2, 98.8  30.3 (13.7, 54.5) 96.1 0.9 0.4, 97.8  

BDI, Beck Depression Inventory; CES-D, Centre for Epidemiological Studies Depression Scale; GAD-7, Generalised Anxiety Disorder Scale; GBTM, Group-Based Trajectory Modelling; GMM, Growth Mixture Models; HADS-A, Hospital Anxiety and Depression Scale-Anxiety Subscale; HADS-D, Hospital Anxiety and Depression Scale-Depression Subscale; HIV, human immunodeficiency virus; LCGA, Latent Class Growth Analysis; PHQ-9, Patient Health Questionnaire; SCI, spinal cord injury; STAI, The State-Trait Anxiety Inventory; TBI, traumatic brain injury; PI, prediction intervals.

The overall pooled prevalence of participants following a clinical depression trajectory was 11.8% (95% CI: 9.2, 14.8), with a high amount of heterogeneity observed (Q52 = 3,908.9, p < 0.001, I2 = 98.7, τ2 = 0.9). The moderating effect of health condition was significant (Q7 = 81.94, p < 0.001), as was the moderating effect of outcome measure (Q3 = 37.99, p < 0.001), and the study country (Q3 = 15.02, p = 0.002), see Table 1. The moderating effects of recruitment setting (Q2 = 2.44, p = 0.30), type of analysis (Q1 = 3.14, p = 0.08), sample size (B = −0.0002, 95% CI: 0.0001, −0.0005, Q1 = 3.6, p = 0.06), and study year (B = 0.04, 95% CI: −0.03, 1.0, Q1 = 1.14, p = 0.29) were nonsignificant.

Anxiety

There were 27 included studies reporting anxiety trajectories, representing 14,269 participants. Of these 27 studies, 2-trajectory (k = 9) and 3-trajectory (k = 9) solutions were the most commonly identified. A 4-trajectory solution was the second most reported (k = 7), while 6- and 7-trajectory solutions were identified by one study each (see Table S5). Most studies (k = 23, 85%) identified one or more trajectories that remained in the nonclinical range, with between 4% and 92% of participants belonging to these trajectories (see Fig. 3). Of the 27 anxiety studies, most (k = 18, 66%) identified one or more trajectories that remained in the clinical range, with the proportion of participants in these trajectories ranging from 2% to 63%. Furthermore, some studies (k = 12, 44%) identified stable moderate trajectories, in which participant symptom scores were close to the measure’s cut-off. The proportion of participants in these trajectories ranged from 3% to 47%.

Fig. 3.

Depiction of anxiety trajectories.

Fig. 3.

Depiction of anxiety trajectories.

Close modal

Regarding transient trajectories, some studies (k = 5, 19%) reported increasing trajectories, with three reporting trajectories that increased from the nonclinical to clinical range. Furthermore, some studies (k = 11, 40%) reported decreasing symptom trajectories, with 10 that decreased from the clinical to the nonclinical range. Finally, two studies (7%) reported trajectories that followed a fluctuating pattern (e.g., increasing and then decreasing again, or vice versa).

Meta-Analysis

The overall pooled prevalence of participants following a nonclinical anxiety trajectory was 73.4% (95% CI: 66.3, 79.5), with a high amount of heterogeneity observed (Q25 = 1,465.1, p < 0.001, I2 = 98.3, τ2 = 0.7). Subgroup analyses showed that study country was significant (Q2 = 8.0, p = 0.02). The moderating effects of health conditions (cancer or cardiovascular disease; Q1 = 0.9, p = 0.35), recruitment setting (Q2 = 4.48, p = 0.11), outcome measure (Q2 = 5.31, p = 0.07), and sample size (B = 0.003, 95% CI: −0.0003, 0.0008, Q1 = 0.86, p = 0.35) were not significant. See Table 1 for moderator analysis results.

The overall pooled prevalence of participants following a clinical anxiety trajectory was 13.7% (95% CI: 9.3, 19.7), with a high amount of heterogeneity observed (Q25 = 1,111.7, p < 0.001, I2 = 97.8, τ2 = 1.0). The moderating effect of recruitment setting was significant (Q2 = 14.47, p < 0.001), as was the effect of analysis type (Q1 = 9.65, p = 0.01) and the study country (Q2 = 9.0, p = 0.01). Meta-regression also identified a significant association between sample size and clinical trajectory prevalence (B = −0.002, 95% CI: −0.002, −0.0007, Q1 = 12.89, p < 0.001). The moderating effects of health conditions (cancer or cardiovascular disease; Q1 = 0.06, p < 0.80) and outcome measure (Q2 = 4.88, p = 0.09) were nonsignificant.

Sensitivity Analyses Removing Outlier Studies

Regarding the nonclinical depression outcomes, 8 studies were removed as outliers. With this reduced sample, the overall pooled prevalence of participants following a nonclinical depression trajectory was largely unchanged at 72.8% (95% CI: 69.9, 75.5), with a high amount of heterogeneity observed (Q48 = 2,252.8, p < 0.001, I2 = 97.9, τ2 = 0.2). A total of 7 outlying studies were removed from the clinical depression trajectories, resulting in a similar pooled prevalence of 9.3% (95% CI: 7.7, 11.3), with a high amount of heterogeneity observed (Q48 = 1,793.3, p < 0.001, I2 = 97.3, τ2 = 0.5). Moderator analyses yielded a slightly different pattern of results (see Table S2). Regarding nonclinical depression trajectories, the study country was no longer significant. Regarding clinical depression trajectories, the type of chronic condition also became nonsignificant.

Regarding the nonclinical anxiety outcomes, 3 studies were removed as outliers. With this reduced sample, the overall pooled prevalence of participants following a nonclinical anxiety trajectory was 77.9% (95% CI: 72.0, 82.9), and heterogeneity remained high (Q22 = 1,104.4, p < 0.001, I2 = 98.0). A total of 4 outlying studies were removed from the clinical anxiety trajectories, resulting in an overall pooled prevalence of 11.3% (95C CI: 6.5, 18.7) of participants following a clinical anxiety trajectory (again with high heterogeneity; Q21 364.8, p < 0.001, I2 94.2). Recruitment setting became a significant moderator of the prevalence of participants following a nonclinical anxiety trajectory (see Table S6).

Predictors

The individual studies’ approach to reporting and analysing predictors of trajectory membership was highly variable (see Table S7). For example, 25 studies used ANOVA or χ2 approaches, while 14 used binominal logistic regression, and 24 used multinomial logistic regression analyses. We chose to examine and synthesise only predictors relating to disease severity, which were examined in 41 studies. These included any widely used indicators of disease severity, degree of medical control, or prognosis, for example, cancer stage, the model for end-stage liver disease (MELD) score, level of SCI, and blood glucose levels (HbA1c). See online supplementary Table 3 for disease severity predictor results.

Cancer Stage

Of the 12 studies in samples with cancer, cancer stage was examined as a predictor in 10. Across these 10 studies, 7 reported nonsignificant results, while 4 found evidence of an association between cancer stage and trajectory membership. All four of these studies found evidence that participants following a decreasing trajectory of symptoms were more likely to have higher cancer stages. In addition, there was some evidence that participants with higher cancer stages were more likely to belong to more distressed trajectories. For example, Dunn et al. [38] reported that, compared to participants with stages 0, I, and II cancer, those with stages III and IV cancer were 1.8 times more likely to belong to the moderate symptom trajectory, compared to those in the stable low-symptom trajectory.

Liver Disease Severity

Two of the three studies conducted in liver disease examined whether participants’ MELD score was associated with trajectory membership, with neither study finding evidence of an association [39, 40].

Spinal Cord Injury

One of three studies among people with spinal cord injury examined whether the level of SCI (i.e., paraplegia or tetraplegia) was associated with trajectory membership [23]. The relationship was nonsignificant.

Cardiovascular Disease

Of the 16 trajectory studies conducted in cardiovascular disease, predictors were examined in 7. Three studies [37, 41, 42] examined the New York Heart Association Classification, which indicates the severity of heart function limitations [43]. Two of the studies found no association between the classification and trajectory membership. One study found that worse heart functional status was associated with being in the more severe depressive symptom trajectories but was not associated with anxiety trajectory membership [42]. Three studies [42, 44, 45] examined the left ventricular ejection fractions, which indicate the extent of heart damage [46]. No significant results were identified across these studies.

Traumatic Brain Injury

Among the four studies in TBI, all examined predictors of trajectory membership. Two studies examined whether participants’ Glasgow Coma Scale (GCS) scores were associated with trajectory membership, with contradictory results. In one study, more severe TBI, as indicated by a higher GCS score, was associated with a higher probability of belonging to the stable high-depression trajectory than the stable low trajectory, OR = 2.83 (95% CI: 1.02, 7.81), p = 0.04 [47]. However, having a higher GCS score was associated with a higher probability of being in the stable low depressive symptom trajectory compared to the increasing trajectory in another study [48].

Furthermore, one study [49] examined the duration of post-trauma amnesia and identified a positive relationship; patients in the stable high-depression trajectory had a longer post-traumatic amnesia duration than those in the stable low trajectory. Finally, one study [50] examined “time to follow commands” (indicative of the number of days between the date of injury and the date that the patient was able to follow simple motor commands in two successive assessments within a 24-h period). No significant relationship was found between the time to follow commands and trajectory membership.

Diabetes

Five of the 7 studies in diabetes examined predictors, with four examining the association between blood glucose (HbA1c) and trajectory membership. Of these studies, three found significant associations. Two studies found evidence that participants with higher HbA1c were more likely to belong to decreasing trajectories regarding anxiety [51] and depression [52]. In comparison, Kampling et al. [53] found higher HbA1c values among participants following a worsening depressive symptom trajectory (8.2%) compared to those in the no-depression trajectory (7.2%).

Other Health Conditions

A range of other predictors were examined across the health conditions. Two of the three studies that were conducted on participants with Lupus exampled predictors, using various indicators of disease activity. In both studies, no significant relationships were found. In multiple sclerosis, one study [54] identified that a slightly larger proportion of participants with progressive multiple sclerosis belonged to high and moderate trajectories (11% and 30%) compared to those with relapsing-remitting multiple sclerosis (10% and 25%). Furthermore, one study conducted in HIV-positive men and women found that those with an unsuppressed viral load were more likely to be in the stable high-depression trajectory and the moderate fluctuating depression trajectory than the stable low trajectory, respectively [55]. However, another study did not find a significant relationship between detectable viral load and trajectory membership in HIV-positive individuals [22].

The aim of this systematic review and meta-analysis was to synthesise the available studies examining trajectories of depression and anxiety in chronic disease samples. The current review identified 67 studies (n > 60,000) across a broad range of chronic disease groups. Despite diversity in terms of the diseases studied and individual study methods, we observed a high degree of consistency across studies in terms of the number and shape of trajectories. As such, some clear and important findings emerged from the current review.

Perhaps the most noteworthy finding is that most participants with chronic disease follow a trajectory of distress that is low and stable, often termed “resilient” in the broader literature [56]. Meta-analysis results suggest that this is 69% (95% CI: 66, 72) for depression and 73% (95% CI: 66, 80) for anxiety. Interestingly, these results are similar to research examining participants’ adjustment following exposure to trauma, where the average prevalence of people following a resilient trajectory was 66% (95% CI: 62, 70) [56]. From a theoretical standpoint, the current study suggests that most people are able to psychologically adjust to living with chronic diseases, despite the ongoing challenges they may face.

Mirroring the above finding, our results also show that a smaller percentage of participants follow a course of distress that remains consistently high or chronic that is similarly prevalent across anxiety and depressive symptoms; 11% (95% CI: 9, 14) for depression and 14% (95% CI: 9, 17) for anxiety. While participants following these chronic trajectories were usually the minority, these rates still reflect a meaningful subgroup of people. It is important to acknowledge that some of these participants were likely to follow such a trajectory of distress irrespective of the onset of a chronic disease. However, studies examining the trajectories of depression in the general population typically find that between 4% and 7% of participants follow a trajectory of consistently high symptoms [57‒60]. As such, the onset of disease likely precipitates continuous and clinically significant distress for a proportion of people. Consistent with this, controlled studies examining the prevalence of psychiatric disorders show that people with chronic disease are around 1.5 to 1.8 times more likely to report depression or anxiety, respectively, compared with the general population [61]. Within these chronic disease groups, it is important to recognise the complex relationships between comorbid chronic disease, depression and anxiety symptoms, and environmental factors. For example, the onset of chronic disease can often be influenced by environmental challenges which exceed an individual’s ability to cope (i.e., allostatic overload [62‒65]). These relationships can also be mediated by damaging health behaviours, such as excessive alcohol intake [62‒65]. Indeed, longitudinal studies show that stressful life events often predate the onset of chronic disease, and evidence of allostatic overload has been identified in individuals with conditions such as coronary heart disease and fibromyalgia [62, 63].

Compared to the two stable, resilient, and chronic trajectories, transient trajectories were less consistently observed. For depression, approximately half the studies reported increasing and decreasing trajectories (k = 28 and k = 31, respectively). Only 5 of the 27 studies on anxiety identified increasing trajectories, while decreasing trajectories were more common (k = 11 of 27). Developing ways to identify individuals who are likely to remit on their own would allow for more efficient and cost-effective focussing of psychological resources to help individuals who may experience chronic distress or increasing distress. However, because of the heterogeneity in chronic disease and study methodology, it is difficult to understand what factors may be responsible for the different presence of transient trajectories. This is an important avenue for future research.

The current study also found that the nature or severity of disease does not appear to meaningfully influence participants’ trajectory outcomes. While the moderator analyses were statistically significant for depression, visual examination of the pooled prevalence across health conditions suggests that no health condition appears to confer markedly more or less risk of poor adjustment. Except for liver disease, the proportion of people following a resilient trajectory ranged from 60% (in SCI) to 80% (in diabetes). Similarly, and with the same exception of liver disease, the proportion of participants following a chronic trajectory ranged from 6% (in cancer) to 14% (in traumatic brain injury). Regarding participants following a clinical trajectory, health condition became a nonsignificant moderator after removing outlying studies. In anxiety, there were enough studies to compare cancer and cardiovascular disease, where the pooled prevalence of resilient and chronic trajectories was similar.

A similar pattern of findings was observed regarding disease-related predictors of trajectory class membership. In most studies, disease severity predictors were nonsignificant. For example, nonsignificant results were found in the two studies examining the relationship between the level of SCI and trajectory membership. Similarly, MELD status did not predict trajectory membership in both studies examining it as a predictor in liver disease [39, 40]. In some cases, significant associations were found, though they appeared weak. For example, while one study [54] found that MS type predicted trajectory membership, the proportion of participants with primary progressive MS following a chronic trajectory was 11%, while the proportion of participants with relapsing-remitting MS following the same trajectory was 10%. In fact, many studies identifying statistically significant results were in samples with many thousands of participants (like the above), which are statistically powered to detect very small effects and are at risk of Type II errors. One potential exception to this trend was for diabetes, where there was evidence of an association between blood glucose levels and worse depression trajectory membership [51‒53]. Interestingly, prospective studies have shown that depression symptoms may influence blood glucose levels, likely due to their influence on participants’ self-management behaviours [66].

Our study also examined whether a range of methodological factors (e.g., analysis type, sample size, and instrument used) may be associated with study outcomes. The type of analysis used was a significant moderator in studies investigating clinical anxiety trajectories, and this remained significant in sensitivity analyses. A larger proportion of participants belonged to the clinical anxiety trajectory in studies using Latent Class Growth Analysis (LCGA) or Group-Based Trajectory Modelling (GBTM) methods compared to studies using Growth Mixture Modelling. This raises questions about whether the type of analytic method may influence the number of individuals who are categorised as having chronic anxiety. Indeed, LCGA and GBTM usually identify more trajectories than GMM as it assumes that all observations are independent [67]. Therefore, the larger proportion of participants in clinical anxiety trajectories may be due to the LCGA method, and not the severity of their anxiety symptoms.

For depression, the choice of instrument moderated the prevalence of participants belonging to the chronic trajectory. Interestingly, the BDI (which observed the largest proportion of participants in the chronic trajectory; 20%, 95% CI: 15, 27) contains numerous items assessing somatic symptoms of depression that may be confounded by living with chronic diseases (e.g., weight loss, worry about health, fatigue, changed libido). In comparison, a much smaller proportion of participants following the chronic trajectory were observed in the HADS-D (5%, 95% CI: 2, 9), which was specifically developed for use in medical populations. While concerns for the performance of the BDI in medical populations have been raised in previous studies [68], others have found its sensitivity and specificity to remain adequate [69‒71]. While this moderator was not significant for anxiety, there was a similar trend indicating that the choice of instrument may affect study outcomes. For example, the state-trait anxiety inventory (STAI) identified that 22.5% (95% CI: 10.1, 42.9) of participants followed a chronic anxiety trajectory, while the Generalised Anxiety Disorder 7-item Scale (GAD-7) yielded a result of 0.3% (95% CI: 0.0, 18.7; p value for this moderator = 0.09). It is crucial for researchers to remain mindful of the overlapping nature of many depression or anxiety symptoms with those of health conditions or medication side effects. It is also important that future research use instruments that have been developed, or successfully validated, in participants with health conditions.

This study has some noteworthy implications for clinical practice. First, our results suggest that consistently poor adjustment affects 10–15% of people with chronic disease at most, which lasts for an average of 3 years and 9 months (3.76 years), and that this is consistent across illness groups. Indeed, this average duration highlights that a proportion of individuals with comorbid chronic disease and anxiety or depression symptoms may not naturally remit. Therefore, health care systems can estimate that this proportion of patients may require additional psychosocial or other support in order to protect against the adverse outcomes associated with distress in the context of disease. Importantly, health care professionals cannot assume that participants with greater disease severity are more likely to require psychological or psychiatric treatment. Specifically, the results indicate that across diverse chronic diseases (e.g., cancer, spinal cord injury, and diabetes), similar proportions of individuals report chronic clinical and nonclinical trajectories of depression or anxiety. Given adjustment appears less related to the disease and its severity, it is important for clinicians to consider the factors that may explain why certain individuals struggle to psychologically adjust to their illness. Two such factors that warrant future attention include potential roles of allostatic load [62, 63] and specific health behaviours and attitudes [64, 65] in the process of developing and subsequent adjustment to chronic disease. There is considerable literature concerning these factors, yet few longitudinal studies have examined their potential roles in adjustment over time.

The current study also raises important questions for future research. Our results suggest that universally targeted treatments are unlikely to be clinically useful or cost-effective, given the majority of people do not require them in the first place, while others may follow a natural course of adjustment. It would be valuable for future treatment studies to examine whether their outcomes differ depending on the duration and severity of participants’ distress following the onset of chronic disease or to impose a minimum duration and severity as part of their inclusion criteria. While the current study excluded samples where all participants were receiving some form of psychological treatment (e.g., psychotherapy or psychotropic medication), it would be interesting to complement the findings of the current study with insight into trajectories of adjustment within the context of such treatment.

Future research is also needed to clarify the relationship between potential predictors of adjustment trajectories, such as country of residence. Study country significantly moderated the prevalence of all four trajectories, though it was only possible to compare four countries with sufficient data: Australia, China, the Netherlands, and the USA. In general, it appeared that participants from Australia demonstrated more favourable trajectories of adjustment, with higher proportions following a nonclinical trajectory across anxiety and depression and lower proportions following a clinical trajectory. For other countries, results were less clear. For both the Netherlands and the USA, depression trajectories were similar to the overall sample, while anxiety trajectories were less favourable. For example, respectively, 30% and 35% of participants from the USA and Netherlands followed a clinical anxiety trajectory. These differences may reflect sociocultural and structural differences between countries, which vary in terms of factors such as access to health care and mental health care [72‒74] and stigma around mental health [75‒77]. However, it may be that these findings reflect methodological differences between studies that could not be assessed in the current study. Therefore, future studies across countries that utilise similar methods (e.g., the same analyses, outcome measures, and chronic disease populations) will help clarify whether country of residence influences participants’ adjustment to chronic disease.

There would also be significant value in an individual patient data (IPD) meta-analysis of the included studies. While the current meta-analysis results and confidence intervals suggest a degree of between-study consistency, indications of heterogeneity and uncertainty were high and remained high after removing outliers. While high heterogeneity has been observed in other meta-analysis of prevalence [31], an IPD approach would allow researchers to standardise various aspects of methodology (e.g., analytic method) which may reduce study heterogeneity. In addition, IPD meta-analyses would allow researchers to examine the contribution of potentially important participant characteristics (e.g., age and sex) that are not possible with an aggregate approach. As such, an IPD approach may overcome some of the valid shortcomings of meta-analytic approaches, particularly the risk of missing important clinical detail with the use of large aggregate analyses [78].

It is also important to note that a proportion of participants in each of the included studies were likely receiving psychological care. Of the 20 studies that examined and reported this, between 3 and 48% of participants were prescribed with antidepressant medication or receiving psychotherapy. As such, the findings observed include a subsample of participants who have accessed some form of mental health care. Importantly, among studies reporting antidepressant use, it is unclear whether these were prescribed for mood concerns (e.g., antidepressant medication is commonly prescribed to manage chronic pain symptoms). Further research, again using individual patient data techniques, is important to compare the trajectories of adjustment among groups who do versus do not receive concurrent mental health care. With these concerns in mind, it is possible to anticipate the outcome of such comparative analysis. All included studies that examined whether mental health treatments were associated with trajectory membership found that concurrent medication or psychotherapy was associated with more severe symptom trajectories. As such, removing these participants from study samples may lead to a higher prevalence of participants following a resilient trajectory, and a lower prevalence of those following a chronically distressed trajectory.

The current results should be interpreted with some limitations in mind. Firstly, while many other psychosocial, demographic, and health-related variables were examined as potential predictors of trajectory class in the included studies, they were beyond the scope of the current review. Second, we do not know what proportion of participants with a chronic high or increasing trajectory had psychological issues prior to chronic disease onset. Third, the frequency of measurement within the included studies was variable, with many studies assessing individuals annually. Furthermore, some studies began with monthly or bi-monthly assessments before increasing the time between measurements to once a year. Therefore, with most of the assessments spaced far apart, short-term periods of psychological distress that are associated with chronic illness may have been missed. However, a strength of this meta-analysis is that most included studies (k = 41, 61%) measured trajectories for a duration of 2 years or more; thus, we may be able to assume that these trajectories are stable over long periods of time.

In summary, the current study is the first known examination of psychological distress and adjustment trajectories in chronic disease groups. While the results of the study should be interpreted in the context of high heterogeneity and an aggregate data approach, a consistent trend was observed across a broad range of chronic diseases. Specifically, the majority of participants follow a resilient nonclinical trajectory, and a minority follow a chronic distressed trajectory. While resilience in the face of traumatic or stressful life events is well known [56], this has not been as well documented among people living with chronic disease. In addition, the nature and severity of chronic disease appear to have a limited association with trajectory membership. Further research is needed to understand the unique pathways and risk factors that bring about poor psychological adjustment among the many people living with chronic disease.

An ethics statement is not applicable because this study is based exclusively on published literature.

The authors have no conflicts of interest to declare.

This research was enabled by funding from a Macquarie University Research Fellowship (MQRF; A.J.S.) and the Australian National Health and Medical Research Council (NHMRC).

Amelia J. Scott was responsible for conceptualisation and design of the study, registration of the protocol, article search and screening, data collection and curation, statistical analysis, and preparation of the manuscript. Ashleigh B. Correa was responsible for article search and screening, data collection, statistical analysis, and preparation of the manuscript. Madelyne A. Bisby was responsible for data collection and curation and review of the manuscript. Blake F. Dear was responsible for conceptualisation and design of the study and review of the manuscript.

All data generated or analysed during this study are included in this article and its online supplementary material files. Further enquiries can be directed to the corresponding author.

1.
Hajat
C
,
Stein
E
.
The global burden of multiple chronic conditions: a narrative review
.
Prev Med Rep
.
2018 Dec
12
284
93
.
2.
Buchberger
B
,
Huppertz
H
,
Krabbe
L
,
Lux
B
,
Mattivi
JT
,
Siafarikas
A
.
Symptoms of depression and anxiety in youth with type 1 diabetes: a systematic review and meta-analysis
.
Psychoneuroendocrinology
.
2016 Aug
70
70
84
.
3.
Clarke
DM
,
Currie
KC
.
Depression, anxiety and their relationship with chronic diseases: a review of the epidemiology, risk and treatment evidence
.
Med J Aust
.
2009
190
S7
S54
60
.
4.
Lotfaliany
M
,
Bowe
SJ
,
Kowal
P
,
Orellana
L
,
Berk
M
,
Mohebbi
M
.
Depression and chronic diseases: Co-occurrence and communality of risk factors
.
J Affect Disord
.
2018 Dec
241
461
8
.
5.
Walker
ER
,
Druss
BG
.
Cumulative burden of comorbid mental disorders, substance use disorders, chronic medical conditions, and poverty on health among adults in the U.S.A
.
Psychol Health Med
.
2017 Jul
22
6
727
35
.
6.
Holden
L
,
Scuffham
PA
,
Hilton
MF
,
Ware
RS
,
Vecchio
N
,
Whiteford
HA
.
Health-related productivity losses increase when the health condition is co-morbid with psychological distress: findings from a large cross-sectional sample of working Australians
.
BMC Public Health
.
2011 Dec
11
1
417
.
7.
DiMatteo
MR
,
Lepper
HS
,
Croghan
TW
.
Depression is a risk factor for noncompliance with medical treatment: meta-analysis of the effects of anxiety and depression on patient adherence
.
Arch Intern Med
.
2000 Jul
160
14
2101
7
.
8.
Prince
M
,
Patel
V
,
Saxena
S
,
Maj
M
,
Maselko
J
,
Phillips
MR
No health without mental health
.
2007
. p.
370
.
9.
de Ridder
D
,
Geenen
R
,
Kuijer
R
,
van Middendorp
H
.
Psychological adjustment to chronic disease
.
Lancet
.
2008 Jul
372
9634
246
55
.
10.
Dear
BF
,
Scott
AJ
,
Fogliati
R
,
Gandy
M
,
Karin
E
,
Dudeney
J
.
The chronic conditions course: a randomised controlled trial of an internet-Delivered transdiagnostic psychological intervention for people with chronic health conditions
.
Psychother Psychosom
.
2022
;
91
(
4
):
265
76
.
11.
Moss-Morris
R
.
Adjusting to chronic illness: Time for a unified theory
.
Br J Health Psychol
.
2013
;
18
(
4
):
681
6
.
12.
Beltman
MW
,
Voshaar
RCO
,
Speckens
AE
.
Cognitive-behavioural therapy for depression in people with a somatic disease: meta-analysis of randomised controlled trials
.
Br J Psychiatry
.
2010 Jul
197
1
11
9
.
13.
Bidstrup
PE
,
Christensen
J
,
Mertz
BG
,
Rottmann
N
,
Dalton
SO
,
Johansen
C
.
Trajectories of distress, anxiety, and depression among women with breast cancer: looking beyond the mean
.
Acta Oncol
.
2015 May
54
5
789
96
.
14.
Larsen
AM
,
Osborn
L
,
Ronen
K
,
Richardson
BA
,
Jiang
W
,
Chohan
B
.
Trajectories of depression symptoms from pregnancy through 24 months postpartum among kenyan women living with HIV
.
J Acquir Immune Defic Syndr
.
2022 Aug
90
5
473
81
.
15.
Bonanno
GA
.
Loss, trauma, and human resilience: have we underestimated the human capacity to thrive after extremely aversive events
.
Am Psychol
.
2004
;
59
(
1
):
20
8
.
16.
Nagin
DS
.
Group-Based trajectory modeling: an overview
.
Ann Nutr Metab
.
2014
65
2–3
205
10
.
17.
Ram
N
,
Grimm
KJ
.
Growth mixture modeling: a method for identifying differences in longitudinal change among unobserved groups
.
Int J Behav Dev
.
2009 Nov
33
6
565
76
.
18.
Baron
E
,
Bass
J
,
Murray
SM
,
Schneider
M
,
Lund
C
.
A systematic review of growth curve mixture modelling literature investigating trajectories of perinatal depressive symptoms and associated risk factors
.
J Affect Disord
.
2017 Dec
223
194
208
.
19.
Henselmans
I
,
Helgeson
VS
,
Seltman
H
,
de Vries
J
,
Sanderman
R
,
Ranchor
AV
.
Identification and prediction of distress trajectories in the first year after a breast cancer diagnosis
.
Health Psychol
.
2010
;
29
(
2
):
160
8
.
20.
Kong
D
,
Lu
P
,
Solomon
P
,
Shelley
M
.
Gender-based depression trajectories following heart disease onset: significant predictors and health outcomes
.
Aging Ment Health
.
2022 Apr
26
4
754
61
.
21.
Craig
A
,
Tran
Y
,
Guest
R
,
Middleton
J
.
Trajectories of self-Efficacy and depressed mood and their relationship in the first 12 months following spinal cord injury
.
Arch Phys Med Rehabil
.
2019 Mar
100
3
441
7
.
22.
Bengtson
AM
,
Pence
BW
,
Powers
KA
,
Weaver
MA
,
Mimiaga
MJ
,
Gaynes
BN
.
Trajectories of depressive symptoms among a population of HIV-Infected men and women in routine HIV care in the united states
.
AIDS Behav
.
2018 Oct
22
10
3176
87
.
23.
Bombardier
CH
,
Adams
LM
,
Fann
JR
,
Hoffman
JM
.
Depression trajectories during the first year after spinal cord injury
.
Arch Phys Med Rehabil
.
2016a
97
2
196
203
.
24.
Li
H
,
Marsland
AL
,
Conley
YP
,
Sereika
SM
,
Bender
CM
.
Genes involved in the HPA axis and the symptom cluster of fatigue, depressive symptoms, and anxiety in women with breast cancer during 18 months of adjuvant therapy
.
Biol Res Nurs
.
2020
;
22
(
2
):
277
86
.
25.
Chan
ME
,
Arvey
RD
.
Meta-Analysis and the Development of Knowledge
.
Perspect Psychol Sci
.
2012 Jan
7
1
79
92
.
26.
Haidich
AB
.
Meta-analysis in medical research
.
Hippokratia
.
2010 Dec
14
Suppl 1
29
37
.
27.
Moher
D
,
Shamseer
L
,
Clarke
M
,
Ghersi
D
,
Liberati
A
,
Petticrew
M
.
Preferred reporting items for systematic review and meta-analysis protocols (PRISMA-P) 2015 statement
.
Syst Rev
.
2015 Jan
4
1
1
.
28.
Landis
JR
,
Koch
GG
.
The measurement of observer agreement for categorical data
.
Biometrics
.
1977
;
33
(
1
):
159
74
.
29.
Higgins
JPT
,
Thompson
SG
,
Deeks
JJ
,
Altman
DG
.
Measuring inconsistency in meta-analyses
.
BMJ
.
2003 Sep
327
7414
557
60
.
30.
Ioannidis
JPA
,
Patsopoulos
NA
,
Evangelou
E
.
Uncertainty in heterogeneity estimates in meta-analyses
.
BMJ
.
2007 Nov
335
7626
914
6
.
31.
Migliavaca
CB
,
Stein
C
,
Colpani
V
,
Barker
TH
,
Ziegelmann
PK
,
Munn
Z
.
Meta-analysis of prevalence: I2 statistic and how to deal with heterogeneity
.
Res Synth Methods
.
2022
;
13
(
3
):
363
7
.
32.
IntHout
J
,
Ioannidis
JPA
,
Rovers
MM
,
Goeman
JJ
.
Plea for routinely presenting prediction intervals in meta-analysis
.
BMJ Open
.
2016 Jul
6
7
e010247
.
33.
Ayis
SA
,
Ayerbe
L
,
Crichton
SL
,
Rudd
AG
,
Wolfe
CDA
.
The natural history of depression and trajectories of symptoms long term after stroke: The prospective south London stroke register
.
J Affect Disord
.
2016 Apr
194
65
71
.
34.
Schmitz
N
,
Gariépy
G
,
Smith
KJ
,
Malla
A
,
Wang
J
,
Boyer
R
.
The pattern of depressive symptoms in people with type 2 diabetes: a prospective community study
.
J Psychosom Res
.
2013 Feb
74
2
128
34
.
35.
Morris
A
,
Yelin
EH
,
Panopalis
P
,
Julian
L
,
Katz
P
.
Long-term patterns of depression and associations with health and function in a panel study of rheumatoid arthritis
.
J Health Psychol
.
2011
16
4
).
36.
Malhotra
R
,
Chei
CL
,
Menon
E
,
Chow
WL
,
Quah
S
,
Chan
A
.
Short-term trajectories of depressive symptoms in stroke survivors and their family caregivers
.
J Stroke Cerebrovasc Dis
.
2016 Jan
25
1
172
81
.
37.
Kroemeke
A
.
Depressive symptom trajectories over a 6-year period following myocardial infarction: predictive function of cognitive appraisal and coping
.
J Behav Med
.
2016 Apr
39
2
181
91
.
38.
Dunn
J
,
Ng
SK
,
Holland
J
,
Aitken
J
,
Youl
P
,
Baade
PD
.
Trajectories of psychological distress after colorectal cancer
.
Psychooncology
.
2013
;
22
(
8
):
1759
65
.
39.
Annema
C
,
Roodbol
PF
,
Van den Heuvel
ER
,
Metselaar
HJ
,
Van Hoek
B
,
Porte
RJ
.
Trajectories of anxiety and depression in liver transplant candidates during the waiting-list period
.
Br J Health Psychol
.
2017
;
22
(
3
):
481
501
.
40.
DiMartini
A
,
Dew
MA
,
Chaiffetz
D
,
Fitzgerald
MG
,
deVera
ME
,
Fontes
P
.
Early trajectories of depressive symptoms after liver transplantation for alcoholic liver disease predicts long-term survival
.
Am J Transplant
.
2011 Jun
11
6
1287
95
.
41.
Murphy
BM
,
Ludeman
D
,
Elliott
P
,
Judd
F
,
Humphreys
J
,
Edington
J
.
Red flags for persistent or worsening anxiety and depression after an acute cardiac event: a 6-month longitudinal study in regional and rural Australia
.
Eur J Prev Cardiol
.
2014 Sep
21
9
1079
89
.
42.
Murphy
BM
,
Elliott
PC
,
Higgins
RO
,
Le Grande
MR
,
Worcester
MUC
,
Goble
AJ
.
Anxiety and depression after coronary artery bypass graft surgery: most get better, some get worse
.
Eur J Cardiovasc Prev Rehabil
.
2008b
15
4
434
40
.
43.
Raphael
C
,
Briscoe
C
,
Davies
J
,
Ian Whinnett
Z
,
Manisty
C
,
Sutton
R
.
Limitations of the New York Heart Association functional classification system and self-reported walking distances in chronic heart failure
.
Heart
.
2007 Apr
93
4
476
82
.
44.
Kaptein
KI
,
de Jonge
P
,
van den Brink
RHS
,
Korf
J
.
Course of depressive symptoms after myocardial infarction and cardiac prognosis: a latent class analysis
.
Psychosom Med
.
2006 Oct
68
5
662
8
.
45.
Versteeg
H
,
Roest
AM
,
Denollet
J
.
Persistent and fluctuating anxiety levels in the 18 months following acute myocardial infarction: the role of personality
.
Gen Hosp Psychiatry
.
2015 Jan
37
1
1
6
.
46.
Komoriyama
H
,
Omote
K
,
Nagai
T
,
Kato
Y
,
Nagano
N
,
Koyanagawa
K
.
Lower left ventricular ejection fraction and higher serum angiotensin-converting enzyme activity are associated with histopathological diagnosis by endomyocardial biopsy in patients with cardiac sarcoidosis
.
Int J Cardiol
.
2020 Dec
321
113
7
.
47.
Ren
D
,
Fan
J
,
Puccio
AM
,
Okonkwo
DO
,
Beers
SR
,
Conley
Y
.
Group-based trajectory analysis of emotional symptoms among survivors after severe traumatic brain injury
.
J Head Trauma Rehabil
.
2017 Dec
32
6
E29
37
.
48.
Bombardier
CH
,
Hoekstra
T
,
Dikmen
S
,
Fann
JR
.
Depression trajectories during the first year after traumatic brain injury
.
J Neurotrauma
.
2016b
33
23
2115
24
.
49.
Gomez
R
,
Skilbeck
C
,
Thomas
M
,
Slatyer
M
.
Growth mixture modeling of depression symptoms following traumatic brain injury
.
Front Psychol
.
2017
;
8
:
1320
–.
50.
Neumann
D
,
Juengst
SB
,
Bombardier
CH
,
Finn
JA
,
Miles
SR
,
Zhang
Y
.
Anxiety trajectories the first 10 years after a traumatic brain injury (TBI): a TBI model systems study
.
Arch Phys Med Rehabil
.
2022 Nov
103
11
2105
13
.
51.
Whitworth
SR
,
Bruce
DG
,
Starkstein
SE
,
Davis
WA
,
Davis
TME
,
Skinner
TC
.
Depression symptoms are persistent in Type 2 diabetes: risk factors and outcomes of 5-year depression trajectories using latent class growth analysis
.
Diabet Med
.
2017
;
34
(
8
):
1108
15
.
52.
Sutherland
MW
Longitudinal trajectories of depressive symptoms among youth and young adults with type 1 diabetes: predictors and health outcomes
.
2019
.
53.
Kampling
H
,
Petrak
F
,
Farin
E
,
Kulzer
B
,
Herpertz
S
,
Mittag
O
.
Trajectories of depression in adults with newly diagnosed type 1 diabetes: results from the German Multicenter Diabetes Cohort Study
.
Diabetologia
.
2017 Jan
60
1
60
8
.
54.
Gunzler
DD
,
Morris
N
,
Perzynski
A
,
Ontaneda
D
,
Briggs
F
,
Miller
D
.
Heterogeneous depression trajectories in multiple sclerosis patients
.
Mult Scler Relat Disord
.
2016 Sep
9
163
9
.
55.
Kelso-Chichetto
NE
,
Okafor
CN
,
Cook
RL
,
Abraham
AG
,
Bolan
R
,
Plankey
M
.
Association between depressive symptom patterns and clinical profiles among persons living with HIV
.
AIDS Behav
.
2018 May
22
5
1411
22
.
56.
Galatzer-Levy
IR
,
Huang
SH
,
Bonanno
GA
.
Trajectories of resilience and dysfunction following potential trauma: a review and statistical evaluation
.
Clin Psychol Rev
.
2018 Jul
63
41
55
.
57.
Kuchibhatla
MN
,
Fillenbaum
GG
,
Hybels
CF
,
Blazer
DG
.
Trajectory classes of depressive symptoms in a community sample of older adults
.
Acta Psychiatr Scand
.
2012
;
125
(
6
):
492
501
.
58.
Liang
J
,
Xu
X
,
Quiñones
AR
,
Bennett
JM
,
Ye
W
.
Multiple trajectories of depressive symptoms in middle and late life: Racial/ethnic variations
.
Psychol Aging
.
2011
;
26
(
4
):
761
77
.
59.
Lincoln
KD
,
Takeuchi
DT
.
Variation in the trajectories of depressive symptoms: results from the Americans’ Changing Lives Study
.
Biodemography Soc Biol
.
2010 Apr
56
1
24
41
.
60.
Melchior
M
,
Chastang
JF
,
Head
J
,
Goldberg
M
,
Zins
M
,
Nabi
H
.
Socioeconomic position predicts long-term depression trajectory: a 13-year follow-up of the GAZEL cohort study
.
Mol Psychiatry
.
2013 Jan
18
1
112
21
.
61.
Teesson
M
,
Mitchell
PB
,
Deady
M
,
Memedovic
S
,
Slade
T
,
Baillie
A
.
Affective and anxiety disorders and their relationship with chronic physical conditions in australia: findings of the 2007 National Survey of Mental Health and Wellbeing
.
Aust N Z J Psychiatry
.
2011 Nov
45
11
939
46
.
62.
Fava
GA
,
Sonino
N
,
Lucente
M
,
Guidi
J
.
Allostatic load in clinical practice
.
Clin Psychol Sci
.
2023 Mar
11
2
345
56
.
63.
Fava
GA
,
Cosci
F
,
Sonino
N
,
Guidi
J
.
Understanding health attitudes and behavior
.
Am J Med
.
2023 Mar
136
3
252
9
.
64.
Renzaho
AMN
,
Houng
B
,
Oldroyd
J
,
Nicholson
JM
,
D’Esposito
F
,
Oldenburg
B
.
Stressful life events and the onset of chronic diseases among Australian adults: findings from a longitudinal survey
.
Eur J Public Health
.
2014 Feb
24
1
57
62
.
65.
Sahle
BW
,
Chen
W
,
Melaku
YA
,
Akombi
BJ
,
Rawal
LB
,
Renzaho
AMN
.
Association of psychosocial factors with risk of chronic diseases: a nationwide longitudinal study
.
Am J Prev Med
.
2020 Feb
58
2
e39
50
.
66.
Jung
A
,
Du
Y
,
Nübel
J
,
Busch
MA
,
Heidemann
C
,
Scheidt-Nave
C
.
Are depressive symptoms associated with quality of care in diabetes? Findings from a nationwide population-based study
.
BMJ Open Diabetes Res Care
.
2021 Mar
9
1
e001804
.
67.
Fitzmaurice
G
,
Davidian
M
,
Verbeke
G
,
Molenberghs
G
,
Molenberghs
G
Longitudinal data analysis: a handbook of modern statistical methods
London, United Kingdom
CRC Press LLC
2008
. [cited 2023 Mar 15].Available from: http://ebookcentral.proquest.com/lib/mqu/detail.action?docID=359998.
68.
Wedding
U
,
Koch
A
,
Röhrig
B
,
Pientka
L
,
Sauer
H
,
Höffken
K
.
Requestioning depression in patients with cancer: Contribution of somatic and affective symptoms to Beck’s Depression Inventory
.
Ann Oncol
.
2007 Nov
18
11
1875
81
.
69.
Mystakidou
K
,
Tsilika
E
,
Parpa
E
,
Smyrniotis
V
,
Galanos
A
,
Vlahos
L
.
Beck Depression Inventory: exploring its psychometric properties in a palliative care population of advanced cancer patients
.
Eur J Cancer Care
.
2007
;
16
(
3
):
244
50
.
70.
Penley
JA
,
Wiebe
JS
,
Nwosu
A
.
Psychometric properties of the spanish beck depression inventory-II in a medical sample
.
Psychol Assess
.
2003
;
15
(
4
):
569
77
.
71.
Sacco
R
,
Santangelo
G
,
Stamenova
S
,
Bisecco
A
,
Bonavita
S
,
Lavorgna
L
.
Psychometric properties and validity of Beck Depression Inventory II in multiple sclerosis
.
Eur J Neurol
.
2016
;
23
(
4
):
744
50
.
72.
World Health Organization
Regional office for europe, policies EO on HS and
, Kroneman M, Boerma W, van den Berg M, Groenewegen P, et al. Netherlands: health system review. World Health Organization. Regional Office for Europe;
2016
[cited 2023 Jun 22].Available from: https://apps.who.int/iris/handle/10665/330244.
73.
Zieff
G
,
Kerr
ZY
,
Moore
JB
,
Stoner
L
.
Universal healthcare in the United States of America: a healthy debate
.
Medicina
.
2020 Nov
56
11
580
.
74.
Yi
B
.
An overview of the Chinese healthcare system
.
Hepatobiliary Surg Nutr
.
2021 Jan
10
1
93
5
.
75.
Boerema
AM
,
Kleiboer
A
,
Beekman
ATF
,
Van Zoonen
K
,
Dijkshoorn
H
,
Cuijpers
P
.
Determinants of help-seeking behavior in depression: a cross-sectional study
.
BMC Psychiatry
.
2016 Dec
16
1
78
.
76.
Wang
K
,
Link
BG
,
Corrigan
PW
,
Davidson
L
,
Flanagan
E
.
Perceived provider stigma as a predictor of mental health service users’ internalized stigma and disempowerment
.
Psychiatry Res
.
2018 Jan
259
526
31
.
77.
Krendl
AC
,
Pescosolido
BA
.
Countries and cultural differences in the stigma of mental illness: the East–West Divide
.
J Cross Cult Psychol
.
2020 Feb
51
2
149
67
.
78.
Concato
J
,
Horwitz
RI
.
Limited usefulness of meta-analysis for informing patient care
.
Psychother Psychosom
.
2019 Aug
88
5
257
62
.