Skip to main content
NCBI home page
Wiley Open Access Collection logoLink to Wiley Open Access Collection
. 2026 Apr 10;117(4):1432–1448. doi: 10.1111/bjop.70072

Relations among daily symptoms of depression

Meghan E Quinn 1,✉, Mary E Kleinman 1, John P Standring 2, Qimin Liu 2,✉
PMCID: PMC13655993  PMID: 41960730

Abstract

Research has often treated depression as a unitary construct, relying on severity scores or diagnostic thresholds; however, recent studies emphasize that depression is a heterogeneous disorder characterized by dynamic symptom interactions. We aimed to identify unique relations among depressive symptoms when examined longitudinally. We used a 28‐day daily diary design in young adults (N = 363). Three symptom networks, estimated from Bayesian structural equation modelling, identified key symptoms that (1) predicted other symptoms within individuals over time (within‐subject temporal), (2) co‐occurred within the same day (within‐subject contemporaneous) and (3) clustered across individuals (between‐subject). Results revealed that (1) at the within‐subject level, higher levels of sleep disturbance, sad mood, and concentration difficulties predicted higher levels of multiple symptoms the following day, (2) at the within‐subject level, sad mood, anhedonia, and fatigue tended to co‐occur with many other symptoms and (3) at the between‐subject level, individuals with higher levels of anhedonia, anxiety and concentration difficulties tended to experience a broader range of depressive symptoms. These findings underscore the complexity of depressive symptom interactions and highlight potential ways in which depression may manifest. Future research should explore the identified relations to clarify causal relations among symptoms as well as trait‐level vulnerability to symptoms.

Keywords: daily diary, depression, network analysis, symptom dynamics

BACKGROUND

Depression is a highly prevalent disorder associated with significant functional impairment (Kessler et al., 2003, 2005; McClintock et al., 2010; Zhou et al., 2023). Depression is sometimes viewed as a unitary latent construct that causes symptoms to emerge, but may be more accurately viewed as a heterogeneous syndrome (i.e., a collection of unique symptoms; Fried, 2015; Goldberg, 2011; Monroe & Anderson, 2015). This conceptualization of depression as a syndrome suggests that symptoms may occur individually, may signal the occurrence of other symptoms, and may even cause other symptoms to occur. Considering individual symptoms of depression in this way holds the potential to enhance our understanding of depression as a construct as well as how the syndrome of depression unfolds and is maintained.

Most often, depression has been operationalized using diagnostic thresholds or by calculating overall levels of symptom severity. As such, the role of specific depressive symptoms often has been overlooked. Recently, however, research has taken a more granular view and examined relations among symptoms (e.g., Ebrahimi et al., 2021). This research often involves network analysis, in which partial correlations are calculated among all observed symptoms (Borsboom & Cramer, 2013; Robinaugh et al., 2020). These networks may reveal many associations among symptoms and are therefore often summarized by additional measures, including centrality (i.e., the extent to which a symptom is connected to all others within a network). Such research has identified certain symptoms like sadness (Chen et al., 2022; De La Torre‐Luque et al., 2023; Moriarity et al., 2021) that are strongly linked to other symptoms. Implications of these findings go beyond individual symptoms signalling the presence of many other symptoms to the possibility that individual symptoms may cause other symptoms to occur and consequently contribute disproportionately to the occurrence of major depressive episodes.

Most research on symptom relations, however, has relied on cross‐sectional data. This approach does not align with the view that symptoms unfold within individuals and precludes observations of how symptoms co‐occur within individuals over time. This is particularly problematic because cross‐sectional, between‐subject associations are not interchangeable with within‐person associations over time, and these two types of associations may even present in opposite directions. For example, at the between‐subject level, individuals with higher average fatigue may report more concentration difficulties, suggesting a stable association. In contrast, within‐subject patterns could look very different—on days when someone feels more fatigued than usual, concentration difficulties may not be experienced and may even be improved due to reduced activity and fewer cognitive demands. Thus, understanding how symptoms unfold within individuals requires knowledge of temporality among symptoms. In recent years, studies have used longitudinal designs to examine relations among symptoms on time intervals ranging from hours to weeks, often via an experience sampling approach (i.e., asking participants repeatedly to report on their current symptoms) to separate within‐subject from between‐subject symptom relations. Studies that have taken this approach have the potential to generate three types of networks: (1) a within‐subject temporal network that estimates lagged associations among symptoms, (2) a within‐subject contemporaneous network that estimates which symptoms tend to co‐occur at the same time point and (3) a between‐subjects network that estimates which symptoms tend to cluster across individuals (e.g., Ebrahimi et al., 2021; Piccirillo & Rodebaugh, 2022; Quinn et al., 2023).

Our previous study modelled depressive symptom dynamics over 89 weeks, generating these networks (Quinn et al., 2023). In the temporal network, where symptoms predicted others the following week, sad mood emerged as highly connected, driven by predictions from worthlessness/guilt, sleep disturbance and psychomotor symptoms. In the contemporaneous network, which captured same‐week co‐occurrence, sad mood, irritability and anhedonia were among the most central. These findings emphasized that (1) symptom relations differ across temporal scales, and (2) many significant contemporaneous associations may reflect directional effects that unfold more quickly than weekly models can detect—pointing to the need for assessments repeating at shorter time intervals.

Recent studies offer insight into depressive symptom relations at daily, or even shorter, time intervals. A number of studies generating symptom networks using time intervals of hours have focused not solely on depressive symptom relations but related research questions, such as estimating links among symptoms of depression and negative thoughts (Marian & Sava, 2023), links among depressive symptoms and social anxiety symptoms (Piccirillo & Rodebaugh, 2022), and evaluating change in networks of depressive symptoms over time (Nemesure et al., 2024). One study, using a daily time interval, examined five depressive symptoms for 40 days during the COVID‐19 pandemic and found that anhedonia and sad mood were predicted by other symptoms the previous day (Ebrahimi et al., 2021). In their contemporaneous model, anhedonia, sad mood, lethargy and worthlessness were highly connected. While this study highlighted the importance of sad mood and anhedonia, it did not model all symptoms comprising Major Depressive Disorder. A more comprehensive evaluation of depressive symptom relations using a daily timescale would contribute to our understanding of how depressive symptoms unfold within individuals.

The purpose of the present study was to identify various patterns of depressive symptom co‐occurrence to better understand how symptoms of depression occur in relation to each other. To do this, our objectives were to identify (1) symptoms that predict and potentially cause each other from day to day, (2) symptoms that tend to co‐occur within the same day and (3) symptoms that tend to co‐occur within a month, when examined across the entire sample. Taking this nuanced view of depression can further our understanding of depression as a heterogeneous syndrome.

To meet these objectives, we recruited a sample of young adult college students, who tend to have high rates of depression (Ibrahim et al., 2013). Participants were enrolled in a 28‐day daily diary study that assessed depressive symptoms. In addition to the nine DSM‐5 symptoms of depression, we included an item measuring irritability, as irritability was highlighted as a central depressive symptom in our prior study's at‐risk youth sample (Quinn et al., 2023). We also included an item measuring anxiety, given its frequent co‐occurrence with depressive symptoms, despite prior mixed findings in symptom networks (Ebrahimi et al., 2021; Piccirillo & Rodebaugh, 2022). We disaggregated symptom relations into between‐subject, within‐subject temporal and within‐subject contemporaneous effects, and visualized these using networks that capture the strength and, for temporal associations, the direction of symptom connections. We also computed centrality scores to quantify each symptom's relative importance. Based on our prior research utilizing weekly modelling (Quinn et al., 2023), we expected sad mood, anhedonia and irritability to emerge as central in the contemporaneous model and sad mood to show strong in‐strength in the temporal model. All other relations were exploratory, intended to inform future hypothesis generation.

METHOD

Participants

Participants (N = 419) were undergraduate students. In the present study, participants were excluded if they did not complete the daily diary portion of the study (n = 21) or if they completed fewer than seven daily diaries (n = 35).1 These exclusions resulted in a final sample of 363. Participants (Mage = 18.89, SDage = 1.10) identified as 60.6% female, 37.5% male, and 1.9% other, and 1.1% American Indian/Native American/Alaska Native, 20.1% Asian/Asian American, 8.8% Black/African American/African, 11.3% Latino/Hispanic, 1.7% Middle Eastern/Arab, 70.8% White/Caucasian and 0.8% other.

Procedure

This study was approved by the Protection of Human Subjects Committee at the College of William & Mary (Protocol No. PHSC‐2023‐01‐18‐16052‐mequinn). For additional methodological details, see the present study's preregistration: https://osf.io/eptm7/overview. Participants were recruited from an undergraduate research pool at the College of William & Mary, a university in the eastern United States, and received course credit in exchange for participation. Participants were not directly contacted via recruitment techniques; interested participants registered for the study online via Sona Systems. All study procedures were completed online. Informed consent was obtained from all participants. Upon enrolling in the study, participants completed a baseline survey which included measures of demographics and baseline depressive symptom severity. After completing the baseline survey, participants were enrolled in the daily diary portion of the study.

In the daily diary portion of the study, each day for 28 days, participants were asked to complete a brief survey containing questions about depressive symptoms. Participants received the survey each day at 7 A.M. and were asked to respond based on how they were feeling the day prior. This approach was chosen to enhance response rates and ensure that responses reflected symptoms over the entire day.

Measures

Symptoms were drawn from the Clinically Useful Depression Outcome Scale (CUDOS), which is a reliable and valid measure of depression (Zimmerman et al., 2008). The CUDOS items correspond to symptoms of a major depressive episode (American Psychiatric Association, 2013) and include anhedonia, sad mood, sleep disturbance (insomnia or hypersomnia), fatigue, appetite change, concentration difficulties or indecision, psychomotor symptoms (agitation or slowing), suicidal ideation, and worthlessness or guilt. The CUDOS was completed at baseline without modification to characterize the sample by asking participants to report symptom levels for the past week. For the daily diary, the CUDOS instructions were adapted to ask participants to ‘Indicate how often you experienced the following yesterday’. In the daily diaries, the items ‘I felt irritable’ from the Brief Irritability Test (BITe; Holtzman et al., 2015) and ‘I felt nervous anxious, or on edge’ from the Generalized Anxiety Disorder‐7 (GAD‐7; Spitzer et al., 2006) were also included. All items were presented on a five‐point Likert scale ranging from 0 (not at all true) to 4 (almost always true).

Data analysis plan

Consistent with DSM‐5 diagnostic criteria, several depressive symptoms (e.g., appetite change, sleep disturbance, psychomotor symptoms) can manifest in mutually exclusive directions (e.g., poor vs. excessive appetite). To ensure coverage of the full diagnostic spectrum, we retained both possible presentations. However, because these items represent alternate expressions of the same underlying symptom, including them separately would introduce linear dependency into the models and compromise model estimation. We therefore operationalized each composite symptom by taking the maximum value across its respective items, a procedure that captures whether a symptom was present at its highest severity regardless of directionality. This approach allowed us to align measurement with DSM criteria while ensuring statistical identifiability of the model.

We used Bayesian structural equation modelling (BSEM)2 with the brms package in R (Bürkner, 2021) to analyse symptom relations. Given that depressive symptoms were measured on ordinal scales, we conducted ordinal regressions with a logit link function with a flexible threshold, allowing category cut points to vary rather than assuming equal spacing between symptom severity levels. Default non‐informative priors were used to ensure results were data‐driven, making them comparable to maximum likelihood estimation. Each model ran 5000 iterations across two chains, with convergence assessed using the Gelman–Rubin statistic (R‐hat ≤ 1.1). Missing data were imputed with predictive mean matching.

To separate different types of symptom relations, we first computed each participant's average symptom levels across time (between‐subject effects) and then calculated daily deviations from these averages (within‐subject effects). To account for the nested data structure, we specified a random intercepts and slopes model, allowing the intercepts and temporal effects to vary across individuals. Each symptom at day d was regressed on all person‐centred symptoms from day d − 1, capturing within‐subject temporal relations, as well as person means of symptoms (except for the outcome) across 28 days, capturing between‐subject associations. By modelling between‐person differences at the level of individual symptoms rather than a single baseline severity score, overall symptom burden is controlled for in a granular manner while preserving heterogeneity in symptom severity. The within‐subject temporal relations examine whether the change in a person's typical symptom levels predicted symptom levels 1 day later, after accounting for trait‐level differences.

To estimate contemporaneous effects, we first computed pseudoresiduals for each symptom at all‐time points. These pseudoresiduals were defined as the observed symptom score (Y) minus the expected value based on the estimated probabilities from the ordinal regression model. Specifically, for a symptom with possible ratings k, the pseudoresidual was computed as Y−PrY^=1×1−PrY^=2×2−PrY^=3×3 where PrY^=k represents the estimated probability of a given symptom having severity k, derived from the BSEM model with between‐subject and within‐subject temporal effects. Using these pseudoresiduals, we specified a random‐intercept BSEM, regressing the pseudoresiduals of each symptom at day d onto all other person‐centred symptoms at the same time point d to estimate contemporaneous relationships. This approach allowed us to assess same‐day symptom associations while controlling for both temporal (lagged) effects and stable between‐subject differences. By computing contemporaneous effects in this way, we ensured that observed symptom co‐occurrences were not confounded by prior symptom levels or individual differences in symptom severity.

Based on coefficients representing between‐subject, temporal, and contemporaneous effects, we constructed variable networks corresponding to Gaussian graphical models using neighbourhood selection procedure (Li & Zhang, 2017; Meinshausen & Bühlmann, 2006; Williams, 2021). In this approach, each symptom is regressed on all other symptoms, and network edges are retained only when the corresponding regression coefficient is credibly different from zero, yielding a sparse representation of conditional associations among symptoms. To control for multiple comparisons across the 11 symptom models, we used 99.55% credible intervals (1–0.05/11), analogous to a Bonferroni correction. This conservative threshold reduces the likelihood of including spurious edges while preserving interpretability of the resulting networks. Networks were plotted using qgraph (Epskamp et al., 2012). The temporal network showed the extent to which each symptom predicted all other symptoms the following day, within subjects (i.e., difference in symptom level from a person's mean level of symptoms predicting symptom levels on the next day). The contemporaneous network showed the strength of the relations among symptoms on the same day, controlling for symptoms the prior week, within subjects (i.e., difference in symptom level from a person's mean level of symptoms predicting symptom levels on the same day). The between‐subjects network showed the strength of relations among mean levels of symptoms over the 28 days, across subjects.

Based on the network edges (i.e., associations among symptoms), we computed strength centralities (i.e., sum of the strength of association from a symptom to each symptom it was connected to) to evaluate symptom importance in terms of the overall strength of their relations to other depressive symptoms (Barrat et al., 2004). For the temporal model, strength was separated into in‐strength (i.e., an estimate of the extent to which symptoms were predicted by other symptoms the previous day) and out‐strength (i.e., an estimate of the extent to which symptoms predicted other symptoms the next day). We computed and plotted centrality using qgraph (Epskamp et al., 2012). Absolute centrality credible interval estimates are included in Table S1. This study's analyses were pre‐registered: https://osf.io/eptm7/?view_only=49cf65ec136a415f8ebec9d3904478bc.

RESULTS

Descriptive statistics

On average, participants completed 81.86% of the 28 daily diaries (M = 22.92, SD = 5.51). See Figure S1 for response rate of each of the 28 days. Across the sample at baseline, depressive symptom severity levels ranged from non‐depressed to severe depression (M = 14.47, SD = 10.61), with 42.70% characterized as non‐depressed, 30.85% as minimal depression, and 26.45% as mild to severe depression. See Table 1 for descriptive statistics for symptoms across the 28 daily diaries.

TABLE 1.

Descriptive statistics for symptoms across 28 daily diaries.

Symptom N Minimum Maximum Mean Standard deviation
Sad or depressed mood 8315 0 4 0.64 0.91
Anhedonia 8307 0 4 0.53 0.88
Appetite change 8320 0 4 0.89 1.12
Sleep disturbance 8320 0 4 1.06 1.21
Psychomotor symptoms 8319 0 4 0.94 1.12
Worthlessness or guilt 8318 0 4 0.64 1.01
Fatigue 8315 0 4 0.97 1.12
Concentration difficulties or indecision 8318 0 4 1.07 1.24
Suicidal ideation 8317 0 4 0.11 0.46
Irritability 8300 0 4 0.64 0.97
Anxiety 8298 0 4 1.09 1.21
Open in a new tab

Note: Survey items could be skipped; as a result, the number of observations varies slightly across symptoms.

Within‐subject networks

Temporal network

Model estimates are included in Table S2. Significant relations among symptoms are depicted in Figure 1. The temporal network illustrates that each symptom significantly predicted itself the following day, suggesting some level of stability in symptom levels from day to day. Neither irritability nor appetite change predicted other symptoms on subsequent days nor were they predicted by any other symptom on the previous day.

FIGURE 1.

FIGURE 1

Open in a new tab

Temporal symptom network. ANH, anhedonia; ANX, anxious mood; APP, appetite change; CDI, concentration difficulties or indecision; FTG, fatigue; IRT, irritable mood; PMO, psychomotor symptoms; SAD, sad or depressed mood; SLP, sleep disturbance; SUI, suicidal ideation; WOG, worthlessness or guilt. Double‐headed arrows represent autocorrelations. Solid green arrows represent positive relations. See the online article for the colour version of this figure.

Although there were several significant relations revealed by the temporal network, we focus our results on symptoms that are most strongly connected to other symptoms. Sleep disturbance, sad or depressed mood, and concentration difficulties or indecision strongly predicted other symptoms on the subsequent day (i.e., they each had high out‐strength centrality; see Figure 2). Higher than typical level of sleep disturbance for individuals predicted more anhedonia, worthlessness or guilt, concentration difficulties or indecision, psychomotor symptoms, and fatigue the following day. Higher than typical level of sad or depressed mood for individuals predicted more suicidal ideation and worthlessness or guilt the following day. Higher than typical level of concentration difficulties or indecision predicted more worthlessness or guilt, psychomotor symptoms, and anxiety the following day.

FIGURE 2.

FIGURE 2

Open in a new tab

Centrality estimates for temporal network. Strength centrality is min‐max normalized to a scale from 0 to 1, with 1 representing the most central symptom relative to the others in the model.

In the temporal model, psychomotor symptoms and worthlessness or guilt were strongly predicted by other symptoms on the prior day (i.e., had high in‐strength centrality; see Figure 2). Psychomotor symptoms were predicted by fatigue, anxiety, sleep disturbance, and concentration difficulties or indecision. Worthlessness or guilt was predicted by sleep disturbance, sad or depressed mood, and concentration difficulties or indecision.

We did not find support for our hypothesis that sad or depressed mood would have high in‐strength in the temporal model. The temporal model indicated that sad or depressed mood was not predicted by any other symptoms.

Contemporaneous network

Model estimates are included in Table S3. Significant relations among symptoms are depicted in Figure 3. With the exception of anxiety, all symptoms were positively related to other symptoms on the same day. As predicted, sad or depressed mood and anhedonia had high centrality (i.e., were strongly connected to other symptoms in the network; see Figure 4). Each of these symptoms co‐occurred with several other symptoms, including a strong relation between the two symptoms. Fatigue also had high centrality and demonstrated links with sleep disturbance, psychomotor symptoms, concentration difficulties or indecision, sad or depressed mood, anhedonia, and appetite change. Overall, symptoms related to negative emotion and cognitions (i.e., sad or depressed mood, irritability, worthlessness or guilt, suicidal ideation) tended to show stronger associations with one another, while physical symptoms (i.e., fatigue, psychomotor symptoms, appetite change and sleep disturbance) were closely interconnected. While connections existed across these domains, anhedonia appeared to play a central role in linking emotional, cognitive and physical symptoms.

FIGURE 3.

FIGURE 3

Open in a new tab

Contemporaneous symptom network. ANH, anhedonia; ANX, anxious mood; APP, appetite change; CDI, concentration difficulties or indecision; FTG, fatigue; IRT, irritable mood; PMO, psychomotor symptoms; SAD, sad or depressed mood; SLP, sleep disturbance; SUI, suicidal ideation; WOG, worthlessness or guilt. Solid green lines represent positive relations. See the online article for the colour version of this figure.

FIGURE 4.

FIGURE 4

Open in a new tab

Centrality estimates for contemporaneous and between‐subjects networks. Strength estimates for the contemporaneous model are on the left. Strength estimates for the between‐subjects model are on the right. Strength centrality is min‐max normalized to a scale from 0 to 1, with 1 representing the most central symptom relative to the others in the model.

Between‐subjects network

Model estimates are included in Table S2. Significant relations among symptoms are depicted in Figure 5. Anhedonia, anxiety, sad or depressed mood, and concentration difficulties or indecision had high centrality (see Figure 4).

FIGURE 5.

FIGURE 5

Open in a new tab

Between‐subjects symptom network. ANH, anhedonia; ANX, anxious mood; APP, appetite change; CDI, concentration difficulties or indecision; FTG, fatigue; IRT, irritable mood; PMO, psychomotor symptoms; SAD, sad or depressed mood; SLP, sleep disturbance; SUI, suicidal ideation; WOG, worthlessness or guilt. Solid green lines represent positive relations. See the online article for the colour version of this figure.

DISCUSSION

The purpose of the present study was to uncover patterns of depressive symptom co‐occurrence with the goal of identifying symptoms that signal the presence of or vulnerability to other symptoms. We modelled three different forms of depressive symptom co‐occurrence: (1) symptoms that predict and potentially cause each other from day to day, (2) symptoms that tend to co‐occur within the same day and (3) symptoms that tend to co‐occur on average within a month, when examined across the entire sample. Results of the present study provide a basis for formulating testable models for future research.

Temporal network

The temporal network illustrated that each symptom significantly predicted itself the following day, suggesting some level of stability in symptom levels from day to day. This is consistent with our understanding of depression as a disorder in which people experience heightened symptoms for most of the day for most days (American Psychiatric Association, 2013). At the same time, by observing within‐subject associations among symptoms, it is clear that despite this moderate stability, symptom levels also fluctuated from day to day in ways that were predictable from other symptom levels.

Contrary to our prediction, we found that no symptoms predicted sad or depressed mood on the following day. This is inconsistent with our previous observation of high in‐strength centrality of sad or depressed mood, though this was found using a weekly time interval (Quinn et al., 2023). Whereas weekly fluctuations in sad or depressed mood may result from other symptoms, daily fluctuations in sad or depressed mood may be the result of factors other than other symptoms of depression such as rumination or other forms of dysfunctional emotion regulation (Connolly & Alloy, 2017).

Psychomotor symptoms were strongly predicted by fatigue, anxiety, sleep disturbance and concentration difficulties or indecision on the prior day. It is worth noting that psychomotor symptoms included psychomotor retardation and agitation. Although our measures captured these as separate items, we combined them into one item in the current study to reduce model complexity and because of limited power to reliably estimate their unique effects across time. However, this decision may obscure distinct temporal relationships. For example, we cannot ascertain whether the link between anxiety and psychomotor symptoms is the result of psychomotor agitation specifically. Future studies should disentangle such effects to better understand how psychomotor symptoms relate to other symptoms of depression.

Higher than typical levels of sleep disturbance for individuals predicted more anhedonia, worthlessness or guilt, concentration difficulties or indecision, psychomotor symptoms, anxiety, and fatigue the following day. This is consistent with findings from longitudinal studies employing a longer timescale suggesting that sleep disturbance may indicate risk for depression (for a review, see Zhang et al., 2022). This result is also consistent with our previous work modelling weekly depressive symptoms, which showed that sleep disturbance had high out‐strength and specifically predicted higher levels of sad mood and fatigue in the following week (Quinn et al., 2023). Further, given the number of symptoms predicted by sleep disturbance in the present study, it is possible that repeated sleep disturbance could give rise to symptoms sufficient for a depressive episode. A causal role of sleep disturbance in depression is consistent with demonstrations of Cognitive Behavioural Therapy for Insomnia alleviating or preventing depression (Boland et al., 2023; Cunningham & Shapiro, 2018). Together, this body of research points to sleep disturbance as a potential causal factor in depression and improvements in sleep as a viable intervention target.

Higher than typical levels of sad or depressed mood for individuals predicted more suicidal ideation and worthlessness or guilt the following day. At a minimum, this indicates that higher levels of sad or depressed mood precede higher levels of these negative forms of thinking. These findings align with both cognitive models of depression and the Three‐Step Theory (3ST) of suicide (Klonsky & May, 2015). Cognitive models suggest that sad mood activates negative schemas and maladaptive thought patterns—such as worthlessness and guilt—that can spiral into more severe cognitive symptoms (Alloy, 1988; Beck, 1979). The 3ST proposes that suicidal ideation arises when individuals experience intense psychological pain and hopelessness; persistent sad mood may contribute to this emotional pain and reinforce a sense that relief is unattainable (Klonsky & May, 2015). Together, these theories suggest that intervening on sad mood may help interrupt the cascade leading to suicidality.

In the temporal model, higher than typical levels of concentration difficulties or indecision predicted more worthlessness or guilt, psychomotor symptoms, and anxiety the following day. This finding also aligns with cognitive theories of depression, which suggest that impairments in cognitive functioning—such as difficulties with concentration and decision‐making—can exacerbate negative self‐perceptions and emotional distress (Beck, 1979; Carriere et al., 2023; Payne & Schnapp, 2014). Specifically, when individuals struggle to focus or make decisions, they may perceive themselves as inadequate or incapable, reinforcing feelings of worthlessness or excessive guilt. These negative self‐assessments can lead to heightened self‐criticism and rumination, further entrenching depressive symptoms (Fernández‐Jiménez et al., 2024; Löw et al., 2020; Stange et al., 2014). Additionally, cognitive impairments can interfere with daily functioning and problem‐solving abilities, contributing to increased anxiety as individuals feel less equipped to manage stressors (Halse et al., 2022; Hughes et al., 2015). Ultimately, these symptom relations suggest that cognitive impairments may contribute to the maintenance and intensification of other depressive symptoms.

Contemporaneous network

The contemporaneous network showed the strength of within‐subject relations among symptoms on the same day, controlling for symptoms the prior day. As such, this model depicts which symptoms tend to co‐occur on the same day. Although correlational and non‐directional, associations depicted in this model may represent causal relations among symptoms that occur on a time lag <1 day. These associations may also represent symptoms that have a shared cause and therefore tend to co‐occur.

As predicted, sad or depressed mood and anhedonia were strongly connected to other symptoms in the network. Each of these symptoms co‐occurred with several other symptoms, including a strong relation between the two symptoms. Our finding that sad or depressed mood and anhedonia were highly connected to other symptoms is consistent with results of contemporaneous models in our previous study using weekly data in a sample of youth at risk for depression (Quinn et al., 2023) and in a study examining symptoms and other health factors (Ebrahimi et al., 2021). Previous networks using cross‐sectional data have also identified sad mood (Beard et al., 2016; Gijzen et al., 2021; Hakulinen et al., 2020; McElroy et al., 2018) and anhedonia (Bringmann et al., 2015; Hakulinen et al., 2020) as central symptoms. Such results cohere with the requirement of the presence of sad mood or anhedonia in the DSM‐5 depression diagnosis (American Psychiatric Association, 2013).

Contrary to our prediction, irritability was not strongly linked to other symptoms in the contemporaneous model. This prediction was based on our previous study, which found that irritability was strongly connected to other symptoms in a contemporaneous model when symptoms were measured weekly (Quinn et al., 2023). A reason for this discrepancy may be that while sad or depressed mood and irritability may co‐occur within the same week, they are less likely to co‐occur on a daily basis. Differences in ages in the two samples may also explain the discrepant results. The previous study included youth aged 7–17 and the present study included young adults aged 18–25. According to the DSM‐5 (American Psychiatric Association, 2013), irritability is not a symptom of depression in adults but is a core symptom alongside depressed mood and anhedonia in children. Thus, although research suggests that irritability does occur in adult depression (Fava et al., 1997, 2010; Fraguas et al., 2007; Jha et al., 2021; Liu & Cole, 2021; Perlis et al., 2009), results of the present study do not support irritability as a central symptom of depression in young adults.

In the contemporaneous model, fatigue also had high centrality and demonstrated links with sleep disturbance, psychomotor symptoms, concentration difficulties or indecision, sad or depressed mood, appetite change, and anhedonia. This is consistent with another study examining daily symptoms of depression and other health factors (Ebrahimi et al., 2021). The co‐occurrence of fatigue and other symptoms of depression is also consistent with the co‐occurrence of high levels of a range of fatigue symptoms (e.g., physical, neurovegetative) and high levels of depressive symptoms (Corfield et al., 2016). Together, studies indicate that fatigue is closely tied to other depressive symptoms and is not as likely to occur independent of other symptoms.

In the contemporaneous model, anxiety was not associated with any other symptoms. This is consistent with other studies that examined anxiety alongside depressive symptoms and found that anxiety was not a central symptom (Ebrahimi et al., 2021; Piccirillo & Rodebaugh, 2022). This is, however, somewhat surprising given high comorbidity between depression and anxiety disorders (Mihić et al., 2024; Saha et al., 2021; Sánchez Hernández et al., 2023). Although the associated disorders are comorbid, it may be that depressive symptoms and the subjective experience of anxiety do not necessarily co‐occur on the same days.

In our analysis, we observed that symptoms associated with negative emotions and negative cognitions (i.e., sad or depressed mood, irritability, worthlessness or guilt, and suicidal ideation) were strongly interconnected, while physical symptoms (i.e., fatigue, psychomotor symptoms, appetite change and sleep disturbance) exhibited robust associations among themselves. Notably, anhedonia—a diminished ability to experience pleasure—emerged as a pivotal symptom that bridges these domains. This suggests that anhedonia may serve as a central node in the symptom network of depression, potentially facilitating the interplay between emotional, cognitive and physical symptoms. Identifying such central symptoms is crucial, as they can act as primary targets for interventions. For example, addressing anhedonia may disrupt the reinforcing cycles among various depressive symptoms, thereby enhancing overall treatment efficacy. Furthermore, the prominence of anhedonia in connecting diverse symptom domains implies that it might either be a common consequence of these interconnected symptoms or share underlying etiological factors with them. Understanding these relationships could provide insights into the pathophysiology of depression and inform more nuanced intervention strategies.

Between‐subjects network

The between‐subjects network showed the strength of the relations among mean levels of symptoms over the 28 days, across subjects. Because this is a between‐subjects network, results are based on individual averages and are therefore irrespective of the timing of the daily symptoms. In other words, this model indicates which symptoms tended to co‐occur in individuals at any point in the study period. In this model, anhedonia, anxiety, sad or depressed mood, fatigue, and concentration difficulties or indecision were strongly connected to other symptoms. This means that individuals who had higher levels of these symptoms tended to have higher levels of other symptoms, relative to others in the sample. It is notable that sad or depressed mood and anhedonia again emerged as central symptoms in the between‐subjects network. This suggests that individuals who report higher average levels of these symptoms across the study period are also more likely to experience a broader range of depressive symptoms. Unlike the contemporaneous model, which captures how symptoms fluctuate together within individuals over time, the between‐subjects model reflects stable individual differences. In this context, the centrality of sad mood and anhedonia may indicate that these symptoms represent trait‐like vulnerabilities—persistent affective states that increase one's overall propensity to experience other depressive symptoms. These findings reinforce the idea that sad mood and anhedonia are cardinal features of depression. Importantly, they may not consistently co‐occur with other symptoms on a moment‐to‐moment or day‐to‐day basis but instead reflect an enduring risk profile that shapes the broader symptom experience across time.

Limitations and future directions

The present study included a sample of young adult college students. Although there were a range of symptoms reported, including clinical levels of depression, samples with various levels of depressive symptoms should be examined to see if they reveal similar results. For example, it is possible that relations observed in the present study would replicate in other samples with relatively low levels of depression, but other relations may be more prominent in a sample with higher levels of depressive symptoms. Relatedly, this sample included people with any level of depressive symptoms and likely included those with and without a history of depression. It would be interesting to observe relations in those with no history of depression and those with a history of depression to examine specific questions about initial onset of depression and recurrence of depression, respectively. In addition, given that the sample of the present study was drawn from a single university in the United States, and consisted of mostly affluent young adults, these results may not generalize to other groups. Symptom relations should be observed in samples with broader representations of demographic characteristics, including age and race.

As noted earlier, we combined opposing symptoms (e.g., appetite loss and increase) into single variables, which allowed us to align our measurement with DSM‐5 criteria while ensuring statistical identifiability of the model. This decision necessarily obscures differences between opposite behavioural manifestations of certain symptoms. Future work with larger samples could model these symptom directions separately to test whether their temporal dynamics diverge.

Although the present study included anxiety and irritability, the focus was otherwise limited to DSM‐5 symptoms of depression. There are many other factors that relate to symptoms in a dynamic way that could be modelled alongside symptoms to further understand depression and inform future research.

CONCLUSION

The present study adds to the growing body of evidence that depressive symptoms co‐occur in a variety of ways, suggesting that symptoms occur in a dynamic fashion. In the present study, we uncovered a number of symptom associations that can be used as a basis for formulating testable, causal models. At a more basic level, the present study indicates that symptoms of depression do not necessarily onset together, which emphasizes the importance of evaluating depressive symptoms within individuals over time. Additionally, when viewing the three models from the present study alongside other models generated in studies using other time intervals, it becomes apparent that understanding co‐occurrence and potential causal relations among depressive symptoms requires modelling symptom relations using a variety of time intervals.

AUTHOR CONTRIBUTIONS

Meghan E. Quinn: Conceptualization; methodology; investigation; data curation; funding acquisition; project administration; writing – original draft; writing – review and editing. Mary E. Kleinman: Data curation; investigation; project administration; writing – original draft; writing – review and editing. John P. Standring: Writing – review and editing; validation; visualization. Qimin Liu: Conceptualization; formal analysis; software; project administration; methodology; visualization; writing – review and editing.

Supporting information

Figure S1

Tables S1–S3

BJOP-117-1432-s001.docx (82.4KB, docx)

ACKNOWLEDGEMENTS

This work was supported in part by a William & Mary Faculty Research Grant.

Footnotes

1

In EMA meta‐research, typical study durations are short (median ≈7 days) and adherence declines as duration increases; thus a 1‐week minimum within a 28‐day protocol is a pragmatic compromise between data quality and retention. In multilevel/Bayesian models, more records from a participant naturally yield more precise participant‐level estimates; at the population level, each observation contributes to the likelihood (i.e., people with more days do contribute more information).

2

Our analytic framework, a Bayesian structural equation model (BSEM) with ordinal vector autoregressive (VAR) structure, shares close similarities with other intensive longitudinal modelling approaches. DSEM (Asparouhov et al., 2018) implements a multilevel VAR within an SEM framework, typically in Mplus, with latent variable centring and dynamic residual structures. Mixed graphical models (Haslbeck & Waldorp, 2020) estimate sparse VAR and contemporaneous networks using penalized likelihood, but generally assume observed continuous or categorical variables without explicit ordinal link functions. Time‐series dynamic latent variable models (Epskamp, 2020) also rely on state‐space formulations of latent dynamics, often estimated via maximum likelihood. Conceptually, these approaches converge on similar decompositions of within‐ and between‐subject dynamics. Our choice of BSEM with ordinal VAR provides (a) direct modelling of ordinal symptom items via threshold models, (b) a fully Bayesian treatment of uncertainty and (c) flexible random effects estimation.

Contributor Information

Meghan E. Quinn, Email: mequinn@wm.edu.

Qimin Liu, Email: qiminliu@bu.edu.

DATA AVAILABILITY STATEMENT

All data and code are available on OSF: https://osf.io/2x9b7/overview. This study was pre‐registered: https://osf.io/eptm7/overview.

REFERENCES

  1. Alloy, L. B. (1988). Cognitive processes in depression. Guilford Press. [Google Scholar]
  2. American Psychiatric Association (Ed.). (2013). Diagnostic and statistical manual of mental disorders (5th ed.). American Psychiatric Association. 10.1176/appi.books.9780890425596 [DOI] [Google Scholar]
  3. Asparouhov, T. , Hamaker, E. L. , & Muthén, B. (2018). Dynamic structural equation models. Structural Equation Modeling: A Multidisciplinary Journal, 25(3), 359–388. 10.1080/10705511.2017.1406803 [DOI] [Google Scholar]
  4. Barrat, A. , Barthélemy, M. , Pastor‐Satorras, R. , & Vespignani, A. (2004). The architecture of complex weighted networks. Proceedings of the National Academy of Sciences, 101(11), 3747–3752. 10.1073/pnas.0400087101 [DOI] [PMC free article] [PubMed] [Google Scholar]
  5. Beard, C. , Millner, A. J. , Forgeard, M. J. C. , Fried, E. I. , Hsu, K. J. , Treadway, M. T. , Leonard, C. V. , Kertz, S. J. , & Björgvinsson, T. (2016). Network analysis of depression and anxiety symptom relationships in a psychiatric sample. Psychological Medicine, 46(16), 3359–3369. 10.1017/S0033291716002300 [DOI] [PMC free article] [PubMed] [Google Scholar]
  6. Beck, A. T. (1979). Cognitive therapy of depression. Guilford Press. [Google Scholar]
  7. Boland, E. M. , Goldschmied, J. R. , & Gehrman, P. R. (2023). Does insomnia treatment prevent depression? Sleep, 46(6), zsad104. 10.1093/sleep/zsad104 [DOI] [PMC free article] [PubMed] [Google Scholar]
  8. Borsboom, D. , & Cramer, A. O. J. (2013). Network analysis: An integrative approach to the structure of psychopathology. Annual Review of Clinical Psychology, 9(1), 91–121. 10.1146/annurev-clinpsy-050212-185608 [DOI] [PubMed] [Google Scholar]
  9. Bringmann, L. F. , Lemmens, L. H. J. M. , Huibers, M. J. H. , Borsboom, D. , & Tuerlinckx, F. (2015). Revealing the dynamic network structure of the Beck depression inventory‐II. Psychological Medicine, 45(4), 747–757. 10.1017/S0033291714001809 [DOI] [PubMed] [Google Scholar]
  10. Bürkner, P.‐C. (2021). Bayesian item response modeling in R with brms and Stan . Journal of Statistical Software, 100(5), 1–54. 10.18637/jss.v100.i05 [DOI] [Google Scholar]
  11. Carriere, J. S. A. , Nelson, A. L. , Cheyne, J. A. , & Smilek, D. (2023). Influences of inattention on perceived self‐efficacy, stress, and depression. Frontiers in Cognition, 2, 1125197. 10.3389/fcogn.2023.1125197 [DOI] [Google Scholar]
  12. Chen, S. , Bi, K. , Lyu, S. , Sun, P. , & Bonanno, G. A. (2022). Depression and PTSD in the aftermath of strict COVID‐19 lockdowns: A cross‐sectional and longitudinal network analysis. European Journal of Psychotraumatology, 13(2), 2115635. 10.1080/20008066.2022.2115635 [DOI] [PMC free article] [PubMed] [Google Scholar]
  13. Connolly, S. L. , & Alloy, L. B. (2017). Rumination interacts with life stress to predict depressive symptoms: An ecological momentary assessment study. Behaviour Research and Therapy, 97, 86–95. 10.1016/j.brat.2017.07.006 [DOI] [PMC free article] [PubMed] [Google Scholar]
  14. Corfield, E. C. , Martin, N. G. , & Nyholt, D. R. (2016). Co‐occurrence and symptomatology of fatigue and depression. Comprehensive Psychiatry, 71, 1–10. 10.1016/j.comppsych.2016.08.004 [DOI] [PubMed] [Google Scholar]
  15. Cunningham, J. E. A. , & Shapiro, C. M. (2018). Cognitive Behavioural therapy for insomnia (CBT‐I) to treat depression: A systematic review. Journal of Psychosomatic Research, 106, 1–12. 10.1016/j.jpsychores.2017.12.012 [DOI] [PubMed] [Google Scholar]
  16. De La Torre‐Luque, A. , Pemau, A. , Galvez‐Merlin, A. , & Garcia‐Ramos, A. (2023). Immunometabolic alterations in older adults with heightened depressive symptom trajectories: A network approach. Aging & Mental Health, 27(11), 2229–2237. 10.1080/13607863.2023.2227114 [DOI] [PubMed] [Google Scholar]
  17. Ebrahimi, O. V. , Burger, J. , Hoffart, A. , & Johnson, S. U. (2021). Within‐ and across‐day patterns of interplay between depressive symptoms and related psychopathological processes: A dynamic network approach during the COVID‐19 pandemic. BMC Medicine, 19(1), 317. 10.1186/s12916-021-02179-y [DOI] [PMC free article] [PubMed] [Google Scholar]
  18. Epskamp, S. (2020). Psychometric network models from time‐series and panel data. Psychometrika, 85(1), 206–231. 10.1007/s11336-020-09697-3 [DOI] [PMC free article] [PubMed] [Google Scholar]
  19. Epskamp, S. , Cramer, A. O. J. , Waldorp, L. J. , Schmittmann, V. D. , & Borsboom, D. (2012). Qgraph: Network visualizations of relationships in psychometric data. Journal of Statistical Software, 48(4), 1–18. 10.18637/jss.v048.i04 [DOI] [Google Scholar]
  20. Fava, M. , Hwang, I. , Rush, A. J. , Sampson, N. , Walters, E. E. , & Kessler, R. C. (2010). The importance of irritability as a symptom of major depressive disorder: Results from the National Comorbidity Survey Replication. Molecular Psychiatry, 15(8), 856–867. 10.1038/mp.2009.20 [DOI] [PMC free article] [PubMed] [Google Scholar]
  21. Fava, M. , Nierenberg, A. A. , Quitkin, F. M. , Zisook, S. , Pearlstein, T. , Stone, A. , & Rosenbaum, J. F. (1997). A preliminary study on the efficacy of sertraline and imipramine on anger attacks in atypical depression and dysthymia. Psychopharmacology Bulletin, 33(1), 101–103. [PubMed] [Google Scholar]
  22. Fernández‐Jiménez, C. , Dumitrache, C. G. , Rubio, L. , & Ruiz‐Montero, P. J. (2024). Self‐perceptions of ageing and perceived health status: The mediating role of cognitive functioning and physical activity. Ageing and Society, 44(3), 622–641. 10.1017/S0144686X22000332 [DOI] [Google Scholar]
  23. Fraguas, R. , Iosifescu, D. V. , Bankier, B. , Perlis, R. , Clementi‐Craven, N. , Alpert, J. , & Fava, M. (2007). Major depressive disorder with anger attacks and cardiovascular risk factors. The International Journal of Psychiatry in Medicine, 37(1), 99–111. 10.2190/Y361-2W83-268H-5W7V [DOI] [PubMed] [Google Scholar]
  24. Fried, E. I. (2015). Problematic assumptions have slowed down depression research: Why symptoms, not syndromes are the way forward. Frontiers in Psychology, 6, 309. 10.3389/fpsyg.2015.00309 [DOI] [PMC free article] [PubMed] [Google Scholar]
  25. Gijzen, M. W. M. , Rasing, S. P. A. , Creemers, D. H. M. , Smit, F. , Engels, R. C. M. E. , & De Beurs, D. (2021). Suicide ideation as a symptom of adolescent depression. A network analysis. Journal of Affective Disorders, 278, 68–77. 10.1016/j.jad.2020.09.029 [DOI] [PubMed] [Google Scholar]
  26. Goldberg, D. (2011). The heterogeneity of “major depression”. World Psychiatry, 10(3), 226–228. 10.1002/j.2051-5545.2011.tb00061.x [DOI] [PMC free article] [PubMed] [Google Scholar]
  27. Hakulinen, C. , Fried, E. I. , Pulkki‐Råback, L. , Virtanen, M. , Suvisaari, J. , & Elovainio, M. (2020). Network structure of depression symptomology in participants with and without depressive disorder: The population‐based health 2000–2011 study. Social Psychiatry and Psychiatric Epidemiology, 55(10), 1273–1282. 10.1007/s00127-020-01843-7 [DOI] [PMC free article] [PubMed] [Google Scholar]
  28. Halse, M. , Steinsbekk, S. , Hammar, Å. , & Wichstrøm, L. (2022). Longitudinal relations between impaired executive function and symptoms of psychiatric disorders in childhood. Journal of Child Psychology and Psychiatry, and Allied Disciplines, 63(12), 1574–1582. 10.1111/jcpp.13622 [DOI] [PMC free article] [PubMed] [Google Scholar]
  29. Haslbeck, J. M. B. , & Waldorp, L. J. (2020). mgm: Estimating time‐varying mixed graphical models in high‐dimensional data. Journal of Statistical Software, 93(8), 1–46. 10.18637/jss.v093.i08 [DOI] [Google Scholar]
  30. Holtzman, S. , O'Connor, B. P. , Barata, P. C. , & Stewart, D. E. (2015). The brief irritability test (BITe): A measure of irritability for use among men and women. Assessment, 22(1), 101–115. 10.1177/1073191114533814 [DOI] [PMC free article] [PubMed] [Google Scholar]
  31. Hughes, A. J. , Beier, M. , Hartoonian, N. , Turner, A. P. , Amtmann, D. , & Ehde, D. M. (2015). Self‐efficacy as a longitudinal predictor of perceived cognitive impairment in individuals with multiple sclerosis. Archives of Physical Medicine and Rehabilitation, 96(5), 913–919. 10.1016/j.apmr.2015.01.008 [DOI] [PMC free article] [PubMed] [Google Scholar]
  32. Ibrahim, A. K. , Kelly, S. J. , Adams, C. E. , & Glazebrook, C. (2013). A systematic review of studies of depression prevalence in university students. Journal of Psychiatric Research, 47(3), 391–400. 10.1016/j.jpsychires.2012.11.015 [DOI] [PubMed] [Google Scholar]
  33. Jha, M. K. , Fava, M. , Minhajuddin, A. , Chin Fatt, C. , Mischoulon, D. , Wakhlu, N. , Trombello, J. M. , Cusin, C. , & Trivedi, M. H. (2021). Anger attacks are associated with persistently elevated irritability in MDD: Findings from the EMBARC study. Psychological Medicine, 51(8), 1355–1363. 10.1017/S0033291720000112 [DOI] [PubMed] [Google Scholar]
  34. Kessler, R. C. , Berglund, P. , Demler, O. , Jin, R. , Koretz, D. , Merikangas, K. R. , Rush, A. J. , Walters, E. E. , & Wang, P. S. (2003). The epidemiology of major depressive disorder: Results from the National Comorbidity Survey Replication (NCS‐R). JAMA, 289(23), 3095. 10.1001/jama.289.23.3095 [DOI] [PubMed] [Google Scholar]
  35. Kessler, R. C. , Chiu, W. T. , Demler, O. , & Walters, E. E. (2005). Prevalence, severity, and comorbidity of 12‐month DSM‐IV disorders in the National Comorbidity Survey Replication. Archives of General Psychiatry, 62(6), 617. 10.1001/archpsyc.62.6.617 [DOI] [PMC free article] [PubMed] [Google Scholar]
  36. Klonsky, E. D. , & May, A. M. (2015). The three‐step theory (3ST): A new theory of suicide rooted in the “ideation‐to‐action” framework. International Journal of Cognitive Therapy, 8(2), 114–129. 10.1521/ijct.2015.8.2.114 [DOI] [Google Scholar]
  37. Li, F. , & Zhang, X. (2017). Bayesian lasso with neighborhood regression method for Gaussian graphical model. Acta Mathematicae Applicatae Sinica, English Series, 33(2), 485–496. 10.1007/s10255-017-0676-z [DOI] [Google Scholar]
  38. Liu, Q. , & Cole, D. A. (2021). Aggressive outbursts among adults with major depressive disorder: Results from the collaborative psychiatric epidemiological surveys. Journal of Psychiatric Research, 135, 325–331. 10.1016/j.jpsychires.2021.01.040 [DOI] [PubMed] [Google Scholar]
  39. Löw, C. A. , Schauenburg, H. , & Dinger, U. (2020). Self‐criticism and psychotherapy outcome: A systematic review and meta‐analysis. Clinical Psychology Review, 75, 101808. 10.1016/j.cpr.2019.101808 [DOI] [PubMed] [Google Scholar]
  40. Marian, Ş. , & Sava, F. A. (2023). Identifying central negative thoughts using experience sampling and network analysis. Cognitive Therapy and Research, 47(5), 707–719. 10.1007/s10608-023-10400-w [DOI] [Google Scholar]
  41. McClintock, S. M. , Husain, M. M. , Greer, T. L. , & Cullum, C. M. (2010). Association between depression severity and neurocognitive function in major depressive disorder: A review and synthesis. Neuropsychology, 24(1), 9–34. 10.1037/a0017336 [DOI] [PubMed] [Google Scholar]
  42. McElroy, E. , Fearon, P. , Belsky, J. , Fonagy, P. , & Patalay, P. (2018). Networks of depression and anxiety symptoms across development. Journal of the American Academy of Child & Adolescent Psychiatry, 57(12), 964–973. 10.1016/j.jaac.2018.05.027 [DOI] [PMC free article] [PubMed] [Google Scholar]
  43. Meinshausen, N. , & Bühlmann, P. (2006). High‐dimensional graphs and variable selection with the lasso. The Annals of Statistics, 34(3), 1436–1462. 10.1214/009053606000000281 [DOI] [Google Scholar]
  44. Mihić, L. , Janičić, B. , Marchetti, I. , Novović, Z. , Sica, C. , Bottesi, G. , Belopavlović, R. , & Jakšić, N. (2024). Comorbidity among depression, anxiety and stress symptoms in naturalistic clinical samples: A cross‐cultural network analysis. Clinical Psychology & Psychotherapy, 31(1), e2927. 10.1002/cpp.2927 [DOI] [PubMed] [Google Scholar]
  45. Monroe, S. M. , & Anderson, S. F. (2015). Depression: The shroud of heterogeneity. Current Directions in Psychological Science, 24(3), 227–231. 10.1177/0963721414568342 [DOI] [Google Scholar]
  46. Moriarity, D. P. , Van Borkulo, C. , & Alloy, L. B. (2021). Inflammatory phenotype of depression symptom structure: A network perspective. Brain, Behavior, and Immunity, 93, 35–42. 10.1016/j.bbi.2020.12.005 [DOI] [PMC free article] [PubMed] [Google Scholar]
  47. Nemesure, M. D. , Collins, A. C. , Price, G. D. , Griffin, T. Z. , Pillai, A. , Nepal, S. , Heinz, M. V. , Lekkas, D. , Campbell, A. T. , & Jacobson, N. C. (2024). Depressive symptoms as a heterogeneous and constantly evolving dynamical system: Idiographic depressive symptom networks of rapid symptom changes among persons with major depressive disorder. Journal of Psychopathology and Clinical Science, 133(2), 155–166. 10.1037/abn0000884 [DOI] [PMC free article] [PubMed] [Google Scholar]
  48. Payne, T. W. , & Schnapp, M. A. (2014). The relationship between negative affect and reported cognitive failures. Depression Research and Treatment, 2014, 396195. 10.1155/2014/396195 [DOI] [PMC free article] [PubMed] [Google Scholar]
  49. Perlis, R. H. , Fava, M. , Trivedi, M. H. , Alpert, J. , Luther, J. F. , Wisniewski, S. R. , & John Rush, A. (2009). Irritability is associated with anxiety and greater severity, but not bipolar spectrum features, in major depressive disorder. Acta Psychiatrica Scandinavica, 119(4), 282–289. 10.1111/j.1600-0447.2008.01298.x [DOI] [PMC free article] [PubMed] [Google Scholar]
  50. Piccirillo, M. L. , & Rodebaugh, T. L. (2022). Personalized networks of social anxiety disorder and depression and implications for treatment. Journal of Affective Disorders, 298, 262–276. 10.1016/j.jad.2021.10.034 [DOI] [PMC free article] [PubMed] [Google Scholar]
  51. Quinn, M. E. , Liu, Q. , Cole, D. A. , McCauley, E. , Diamond, G. , & Garber, J. (2023). Relations among symptoms of depression over time in at‐risk youth. Journal of Psychopathology and Clinical Science, 132(8), 961–971. 10.1037/abn0000847 [DOI] [PMC free article] [PubMed] [Google Scholar]
  52. Robinaugh, D. J. , Hoekstra, R. H. A. , Toner, E. R. , & Borsboom, D. (2020). The network approach to psychopathology: A review of the literature 2008–2018 and an agenda for future research. Psychological Medicine, 50(3), 353–366. 10.1017/S0033291719003404 [DOI] [PMC free article] [PubMed] [Google Scholar]
  53. Saha, S. , Lim, C. C. W. , Cannon, D. L. , Burton, L. , Bremner, M. , Cosgrove, P. , Huo, Y. , & J McGrath, J. (2021). Co‐morbidity between mood and anxiety disorders: A systematic review and meta‐analysis. Depression and Anxiety, 38(3), 286–306. 10.1002/da.23113 [DOI] [PMC free article] [PubMed] [Google Scholar]
  54. Sánchez Hernández, M. O. , Carrasco, M. A. , & Holgado‐Tello, F. P. (2023). Anxiety and depression symptoms in Spanish children and adolescents: An exploration of comorbidity from the network perspective. Child Psychiatry & Human Development, 54(3), 736–749. 10.1007/s10578-021-01286-4 [DOI] [PMC free article] [PubMed] [Google Scholar]
  55. Spitzer, R. L. , Kroenke, K. , Williams, J. B. W. , & Löwe, B. (2006). A brief measure for assessing generalized anxiety disorder: The GAD‐7. Archives of Internal Medicine, 166(10), 1092. 10.1001/archinte.166.10.1092 [DOI] [PubMed] [Google Scholar]
  56. Stange, J. P. , Hamilton, J. L. , Abramson, L. Y. , & Alloy, L. B. (2014). A vulnerability‐stress examination of response styles theory in adolescence: Stressors, sex differences, and symptom specificity. Journal of Clinical Child and Adolescent Psychology: The Official Journal for the Society of Clinical Child and Adolescent Psychology, American Psychological Association, Division 53, 43(5), 813–827. 10.1080/15374416.2013.812037 [DOI] [PMC free article] [PubMed] [Google Scholar]
  57. Williams, D. R. (2021). Bayesian estimation for Gaussian graphical models: Structure learning, predictability, and network comparisons. Multivariate Behavioral Research, 56(2), 336–352. 10.1080/00273171.2021.1894412 [DOI] [PubMed] [Google Scholar]
  58. Zhang, M.‐M. , Ma, Y. , Du, L.‐T. , Wang, K. , Li, Z. , Zhu, W. , Sun, Y.‐H. , Lu, L. , Bao, Y.‐P. , & Li, S.‐X. (2022). Sleep disorders and non‐sleep circadian disorders predict depression: A systematic review and meta‐analysis of longitudinal studies. Neuroscience & Biobehavioral Reviews, 134, 104532. 10.1016/j.neubiorev.2022.104532 [DOI] [PubMed] [Google Scholar]
  59. Zhou, J. , Zhou, J. , Feng, L. , Feng, Y. , Xiao, L. , Chen, X. , Yang, J. , & Wang, G. (2023). The associations between depressive symptoms, functional impairment, and quality of life, in patients with major depression: Undirected and Bayesian network analyses. Psychological Medicine, 53(14), 6446–6458. 10.1017/S0033291722003385 [DOI] [PMC free article] [PubMed] [Google Scholar]
  60. Zimmerman, M. , Chelminski, I. , McGlinchey, J. B. , & Posternak, M. A. (2008). A clinically useful depression outcome scale. Comprehensive Psychiatry, 49(2), 131–140. 10.1016/j.comppsych.2007.10.006 [DOI] [PubMed] [Google Scholar]

Associated Data

This section collects any data citations, data availability statements, or supplementary materials included in this article.

Supplementary Materials

Figure S1

Tables S1–S3

BJOP-117-1432-s001.docx (82.4KB, docx)

Data Availability Statement

All data and code are available on OSF: https://osf.io/2x9b7/overview. This study was pre‐registered: https://osf.io/eptm7/overview.

Articles from British Journal of Psychology are provided here courtesy of Wiley

RESOURCES