Which Depressive Symptoms Really Matter in Older Adults? A Landmark Review Maps the Network

Depression in later life has long been treated as a checklist of symptoms, each one weighed equally in a diagnostic tally. But a growing body of research has flipped that logic on its head, treating symptoms not as isolated boxes to be ticked but as nodes in an interconnected web, where some symptoms sit at the center of the network and exert outsized influence over the rest. Now, one of the most comprehensive efforts yet to map that web in people aged 50 and older has arrived, and its findings are both illuminating and cautionary. A systematic review published in Translational Psychiatry, led by Stephanie Ming Yin Wong of The University of Hong Kong and colleagues, combed through decades of network analysis studies to answer a deceptively simple question: which depressive symptoms actually occupy the most central positions in the psychological networks of middle-aged and older adults, and what determines that hierarchy?

The scale of the undertaking is striking. The team searched PubMed, PsycINFO, and Web of Science from their inception through February 15, 2025, adding one more study identified during peer review. The final tally came to 57 studies encompassing 176 distinct symptom networks, built with 17 different depression scales. Each network represents a statistical portrait of how symptoms co-vary within a sample: nodes stand for individual symptoms, and edges represent the statistical associations between them, estimated in most cases through partial correlation or related graphical modeling techniques. Symptoms are then ranked by centrality metrics, most commonly expected influence or bridge centrality, which quantify how strongly a given node connects to the rest of the network or to other communities of symptoms. In theory, the most central symptoms are the most promising intervention targets, because activating them should, in principle, ripple outward through the network.

The headline finding is that depressed mood remains the king of DSM-based depressive symptom networks. Across the reviewed studies, it was the symptom most frequently ranked among the top three central nodes when networks were built exclusively from DSM-defined depressive symptoms. But the picture shifts dramatically when the lens widens. In networks that incorporated comorbid conditions, such as anxiety, or extended sets of risk and protective factors, depressed mood’s rate of appearing among the most central nodes fell from 80.6 percent to 59 percent. That drop is more than a statistical footnote. It suggests that the presumed hierarchy of depressive symptoms is not an intrinsic property of late-life depression itself, but is partly an artifact of what researchers choose to include in the model. When anxiety symptoms enter the picture, the center of gravity moves.

And where does it move? Toward anxiety. The review found that the anxiety symptoms of uncontrollable worry, trouble relaxing, and nervousness showed the highest node and bridge centrality in comorbidity and extended networks. Bridge symptoms are of particular clinical interest because they connect otherwise separate communities of psychopathology, potentially serving as the conduits through which one disorder fuels another. In this case, anxiety symptoms appear to act as the connective tissue linking anxious and depressive processes in older adults. Beyond anxiety, other bridge symptoms surfaced repeatedly: feeling tired in the morning, and loneliness-related items such as lacking companionship, feeling alone, and feeling isolated. These findings resonate with a broader understanding of late-life mental health, in which social disconnection and somatic complaints are not merely side effects of depression but potentially active drivers of the wider symptom web.

Notably, the review also looked beyond the DSM’s boundaries. Non-DSM depressive symptoms, captured by scales that include items not part of formal diagnostic criteria, produced their own central players. Lack of happiness emerged as the most central non-DSM symptom, a finding that speaks to the anhedonic texture of depression in older populations. In many older adults, depression manifests less as overt sadness and more as a pervasive loss of pleasure and engagement, and the network evidence now gives that clinical intuition quantitative backing. For clinicians and researchers alike, this raises the possibility that scales emphasizing positive affect and enjoyment may capture therapeutically meaningful variation that standard diagnostic checklists miss.

Perhaps the most provocative results concern the factors that shape which symptoms rise to centrality. The authors used Firth penalized multivariable logistic regression, a technique suited to modeling rare outcomes with limited predictors, to examine what predicted whether a DSM-based depressive symptom, a non-DSM symptom, or a comorbid symptom or associated factor landed among the top three central nodes. The environment in which the data were collected mattered enormously. Networks constructed during the COVID-19 pandemic were associated with significantly lower odds of DSM-based depressive symptoms being most central, with an adjusted odds ratio of 0.22, and dramatically higher odds of comorbid symptoms or associated factors taking the top spots, with an adjusted odds ratio of 3.41. In other words, the pandemic did not simply raise or lower depression scores; it reshaped the very architecture of the symptom network, pushing contextual and comorbid factors toward the center.

Sample characteristics and measurement choices also left their fingerprints. Networks built from clinical or subclinical samples showed lower odds of comorbid symptoms and associated factors being most central, with an adjusted odds ratio of 0.22, suggesting that in treatment-seeking populations the depressive symptoms themselves dominate the network, whereas in community samples the surrounding web of risk and protective factors looms larger. Meanwhile, studies that adopted DSM-based depression scales showed a trend, though not statistically conclusive, toward DSM-based depressive symptoms being most central across networks, with an adjusted odds ratio of 3.10 whose confidence interval spanned unity. The message is hard to escape: what we measure shapes what we find, and the choice of depression scale is not a neutral technical decision but a determinant of the scientific conclusions drawn from it.

These findings land at a moment of intense soul-searching within the network analysis community. Critics have questioned the stability of centrality metrics, noting that they can be sensitive to sample size, estimation method, and the specific items included in a network. If central symptoms were merely statistical noise, they would make poor intervention targets. This review does not resolve that debate, but it reframes it. By systematically cataloging 176 networks and quantifying the factors that predict which symptoms become central, Wong and colleagues provide the field with something it has lacked: an evidence base for understanding when centrality findings are likely to be robust and when they are likely to be contingent. Substantial heterogeneity, the authors conclude, is the defining feature of this literature, and network type, environmental circumstances at the time of data collection, and scale selection are the principal sources of that variability.

For the rapidly growing population of adults over 50, the clinical implications are tangible. If anxiety symptoms such as uncontrollable worry and trouble relaxing serve as bridges between emotional disorders in later life, then interventions targeting worry and physiological arousal, from cognitive behavioral approaches to mindfulness-based stress reduction, might do double duty, dampening both anxiety and the depressive symptoms downstream of it. If loneliness-related items and morning fatigue function as bridge symptoms, then social connection interventions and attention to sleep and energy regulation could similarly act at network chokepoints. And if lack of happiness is the most central non-DSM symptom, then treatments that actively cultivate positive affect, rather than merely suppressing negative symptoms, deserve a prominent place in geriatric mental health care. None of these possibilities is proven by a systematic review of cross-sectional networks, and the authors are careful to frame centrality as a hypothesis-generating tool rather than a prescription. But the convergence of evidence across dozens of independent studies gives these targets a credibility that no single analysis could.

The review also issues a quiet warning about the field’s collective methodology. With 17 different depression scales in play and no standardization of network composition, comparisons across studies are fraught, and the apparent inconsistency of centrality findings may reflect that fragmentation as much as any true heterogeneity in the phenomenon itself. The authors’ use of an adapted National Institutes of Health Quality Assessment Tool found that only six of the included studies were of strong quality, a sobering indicator that much of the network literature in this population still has methodological room to grow. As network analysis matures from a fashionable technique into a foundational framework for understanding psychopathology, studies like this one serve as both a map of what we know and a checklist of what we must do better. For now, the central lesson stands: in the minds of older adults, depression is not a flat list of symptoms but a shaped landscape, and the shape depends on where you stand, when you look, and what you choose to measure.

Subject of Research: Network analysis of central depressive and comorbid symptoms in adults aged 50 and older

Article Title: Centrality of symptoms in networks of depression and comorbid psychopathology among adults aged 50 years and older: A systematic review and examination of influencing factors

Article References: Wong, S. M. Y., Ng, Z. L. Y., Wang, Y.-L., Yeung, J. O. L., Liu, T., Leung, D. K. Y., Chan, W. C., Lum, T. Y. S., & Wong, G. H. Y. (2026). Centrality of symptoms in networks of depression and comorbid psychopathology among adults aged 50 years and older: A systematic review and examination of influencing factors. Translational Psychiatry. https://doi.org/10.1038/s41398-026-04460-1

Image Credits: AI Generated

DOI: 10.1038/s41398-026-04460-1

Keywords: depression, network analysis, older adults, centrality, anxiety, comorbidity, loneliness, COVID-19, systematic review, late-life depression, DSM, bridge symptoms