The coronavirus disease 2019 (COVID-19) has increased risk for mental health problems in many subpopulations (Hiscott et al., 2020; Xiang et al., 2020). Compared with their younger counterparts, older adults have been more vulnerable to mental health problems during the pandemic (Bafail, 2022; Lebrasseur et al., 2021) due to their increased likelihood for COVID infections (Liu et al., 2020a) and poorer outcomes (Du et al., 2020). To protect older people against infection, many countries explicitly discouraged contacts with older adults during the pandemic (Peckham et al., 2022) and implemented strict safety measures including social distancing, quarantine, and closure of public places for reducing in-person socialization (González-Blanco et al., 2020). Unfortunately, these measures may also trigger worry, fear, and isolation, exacerbating mental health problems within this population (Vahia et al., 2020).

Loneliness, which has been defined as having less social contact than one desires (Peplau, 1982), and depression are common in older adults (Luanaigh and Lawlor, 2008; Singh and Misra, 2009) and appear to have increased significantly during the COVID-19 pandemic. A meta-analysis of 30 studies estimated 28.6 % of adults older than 65 years reported loneliness during the COVID-19 pandemic (Su et al., 2022). Another study conducted in Poland found that 58.83 % of older adults reported moderate loneliness during the pandemic while 19.15 % reported depressive symptoms (Dziedzic et al., 2021). Several longitudinal studies have also found that older adults have experienced increased loneliness and depression during the COVID-19 pandemic (Bäuerle et al., 2020; Krendl and Perry, 2021; McGinty et al., 2020).

Prospective relationships between loneliness and depression have been well-documented. For example, higher levels of loneliness correspond to an increased risk for future depressive symptoms (Cacioppo et al., 2006; Erzen and Çikrikci, 2018; Lee et al., 2021), while depression can also predict increased loneliness through cognition changes (Burholt and Scharf, 2014; Houtjes et al., 2014). Although these longitudinal studies have examined complex and interactive associations of loneliness and depression, results are based on total scores from standard scales or associations at overall syndrome levels. Consequently, interactions between individual symptoms have remained unexplored within longitudinal study designs.

Network analysis is a novel method that can map interactions of specific symptoms within and between particular syndromes (Epskamp et al., 2018). In network approaches, nodes signify symptoms and edges denote associations between pairs of nodes (Burger et al., 2022) such as symptoms of loneliness and depression that can be visualized and consequently explained. Network analysis can also help to identify the most influential nodes within a symptom network and isolate potentially useful targets for prevention and intervention (Robinaugh et al., 2016). Network analysis has been widely used to explore inter-relationships between psychiatric symptoms in different populations before and during this pandemic (Bai et al., 2022; Fried et al., 2020; Skjerdingstad et al., 2021; Zhao et al., 2022). However, most network studies have relied on cross-sectional designs that reflect only discrete snapshots of symptom connections at fixed time points (Pandis, 2014). Based on the premise that psychopathology arises from a network of symptoms that interact over time (Borsboom, 2017), the use of network methods with longitudinal data could increase understanding of how symptoms change and activate other syndromes through interactions with one another over time. To date, longitudinal network analyses have been applied to detect symptom changes in patients with schizophrenia before and after acute antipsychotic medication treatment (Sun et al., 2023), assess depressive symptoms in elderly adults before and after diagnosis of chronic physical diseases (Airaksinen et al., 2020), and track psychiatric symptoms before and during the COVID-19 pandemic (Chen et al., 2022). However, these studies focused on symptom comparisons between only two waves and did not consider depressive symptoms in tandem with loneliness. Furthermore, past network studies of depression and loneliness were limited by cross-sectional study designs (Mullarkey et al., 2019) and single item assessments of loneliness (Li et al., 2021; Mullarkey et al., 2019).

To address these limitations, we examined changes in a depression and loneliness network model of older adults based on data from three waves of the English Longitudinal Study of Ageing (ELSA) conducted between 2016 and 2020. We hypothesized that the structure and central symptoms of the depression-loneliness network model would show changes during the final assessment phase conducted during early stages of the COVID-19 pandemic.