While individual symptom endorsement and the overall summed scores differed, depressive symptoms exhibiting the greatest importance within the networks were highly similar between sexes: suicidal ideation, worthlessness, and psychomotor retardation/agitation had the highest values of strength centrality in both male and female networks. Moreover, the edge connecting suicidal ideation and worthlessness was identified as one of the central edges in both networks. This aligns with Murri et al.’s study,24 which identified death wishes and pessimism as most central in the network of late-life depressive symptoms for both sexes (mean age = 74 years, 59 % female). Our results also support the meta-analytic findings of Cavanagh et al.,2 who identified no sex differences in depressive symptoms with regard to suicidality or psychomotor retardation. Indeed, evidence supports that older adults with depression often present with pathogenic reactions (e.g., suicidal ideation, feelings of worthlessness) instead of the normal intense distress reactions (e.g., sadness and insomnia) frequently observed in younger individuals.13 Likewise, psychomotor retardation/agitation, considered as a variant of primary degenerative dementia, is common in both major and minor late-life depression.25 Therefore, when examining the network of late-life depressive symptoms, it makes sense that we find this unique symptom profile as the central hub, regardless of sex.
However, we found significant sex differences in measures of global strength and network structure. Although not directly comparable, this finding seems to contrast with results of previous research that has examined cross-sex differences in network structures of depressive symptoms in individuals seeking treatment for major depressive disorder.9,12 Van Borkulo et al.11 also conducted a similar study with male and female adults who had experienced a depressive episode and found no sex differences in global strength. Notably, in the current study, older female patients had denser or more tightly connected symptom networks, as global strength was higher in the female group than in the male group. According to the network perspective, a more densely and strongly connected network indicates that an individual feels more “trapped” in the disordered state (i.e., depression) than someone with a less densely connected network.26 Particularly, more strongly connected networks will feature stronger feedback among their symptoms and thus may be related to greater vulnerability to depression and a less positive prognosis.26 In contrast, we found that older male patients had less dense symptom networks or weaker connections between symptoms. Therefore, they might be more resistant to symptom spreading and have protection against developing depression.
Specifically, one-to-one edge strength tests showed that the magnitudes of the following four edges in the female symptom network were significantly stronger than those in the male symptom network: loss of interest–hopelessness, sleep disturbance–low energy/fatigue, loss of interest–concentration difficulty, and worthlessness–concentration difficulty. Associations among these symptoms may help explain the sex disparity in depression rates among older adults. Although typical “female depression” with a specific symptom profile has not been identified in older adults, it is worth noting that these symptoms characterize the “depletion syndrome” more commonly found specifically among older female adults.27 The depletion syndrome’s form differs somewhat across modeling approaches, but its stable features include deficits in self-worth, no interest in pleasurable activities, decreased energy, a general sense of hopelessness, and psychomotor retardation.28,29 In fact, the finding that psychomotor retardation/agitation exerted a unique influence in the female network indirectly supports this argument.
Among the four edges, the edge between loss of interest and hopelessness, which was also identified as one of the central edges of the female symptom network, seems especially important and novel. Suh et al.30 also found this association in their network analysis study with a community sample of Korean adults (age 20-86 years, 51.5% female). Importantly, these two symptoms represent core elements of suicidal ideation in older adults and are key predictors of suicide, equal in power if not more powerful compared to depression alone.31 Thus, these symptoms should be as thoroughly assessed as other core features of late-life depression among older female adults. The finding that the edge between sleep disturbance and low energy/fatigue was more pronounced in female symptom networks than in male symptom networks accords with previous research.2,11 However, in this study, sleep disturbance and low energy/fatigue were identified as peripheral symptoms in the female symptom network based on the centrality test, even though empirical data suggest that their prevalence is high.32 A potential explanation is that such somatic symptoms are prevalent in late life regardless of whether they are caused by depression, medical illnesses, or an interaction of the two.33
The other edges—loss of interest–concentration difficulty and worthlessness–concentration difficulty—appear appropriate for substantive interpretation. It is well established that cognitive functions such as concentration and attention are distorted in late-life depression, and they possess adequate clinical utility for diagnosing depression in older adults.13 However, to our knowledge, previous geriatric studies have not found associations of concentration difficulties with other symptoms such as loss of interest and worthlessness. Accordingly, this finding may inform theoretical considerations about sex differences in depression among older adults. Simultaneously, future studies should examine treatments aiming to deactivate these symptoms. Although not perfectly analogous, Kim34 showed that a program combining physical and recreational activities such as arts and crafts considerably improved cognitive function (self-esteem, creativity, motor function, concentration, memory) and decreased depression levels among older female adults in Korea.
Nevertheless, the present findings should be interpreted with caution. After the PSM, sex differences in participants’ characteristics persisted, especially with respect to living arrangements and education. These factors may account for differences in the symptom networks observed in the present study. For instance, there were more individuals living alone in the female group compared to the male group, which may explain the increased negative cognitive features presenting in the female network.35 Further, female participants tended to have lower levels of education. Education is correlated with the capacity to direct attention, especially among older women36; this may support the results regarding the concentration difficulty node’s higher connection in the symptom network compared to other symptoms.
This study has several limitations. First, we could not incorporate complex design features of KCHS due to the lack of established methods for network models. Accordingly, this sample should not be considered truly representative of community-dwelling Korean older adults in 2019. Instead, researchers must view our results as those emerging from the analysis of an extensive national survey with a diverse group of participants and with generalizability possibly superior to smaller/convenience samples. In the same vein, it should also be noted that the NCT has only recently been proposed as a method for comparing network structures between groups. Thus, such methods should be interpreted with some caution as they gain more support and as other methods of network comparisons emerge. Second, a self-reported method for assessing depressive symptoms among older adults can lead to misinterpretation and biased responses, which may lower the precision of the analyses. Third, associations between symptoms within the network do not infer causal relationships as the current study used cross-sectional data; although a central symptom is likely to have an influence on other nodes, it may be a more efficient target for intervention if associations with other symptoms are directed outward or at least bidirectional. Fourth, we excluded many female subjects during the PSM, which is a primary shortcoming of the PSM methodology. Fifth, while it is not uncommon for depression scales to have minimal content overlap, having different types of content in scales might generate a different psychopathology network. Accordingly, one should be careful when comparing this study’s results with others’ results. Another potential limitation of these analyses is the relatively low mean level of depression symptoms (male = 7.49 ± 3.31, female = 7.70 ± 3.48 after PSM). There is some preliminary evidence that symptom network connectivity, though not necessarily structure, differs according to levels of symptom severity.37 Future studies should examine whether the depressive symptom network structure varies across different levels of depression severity and among patients meeting criteria for depressive syndromes.