Do Network Centrality Predict Overall Depression Symptom Reduction With Lifting of Social Distancing Protocols?

Hypotheses (Hs) H1: In the cross-sectional depression-metacognitive network at T1, the symptoms depressed mood, little interest/pleasure, and self-blame and the metacognitive coping variables rumination/worry, threat monitoring, situational avoidance, thought suppression, and uncontrollability of thoughts will be the most central variables.

H2: Reductions of the most central variables of the cross-sectional network from T1 to T2 will be more related to overall reduction of symptoms in the same time frame than reductions of the less central variables.

Study design and participants The design is a longitudinal observational survey of the general adult Norwegian population during the COVID-19 pandemic. Eligible participants are all individuals of 18 years and above, who are living in Norway and thus experience identical distancing protocols, and who provide informed consent to participate in the study. For this study, only participants reaching the cut-off value (>= 5; Kroenke et al, 2001) for subclinical symptoms on the PHQ-9 at T1 will be included, amounting to 6 508 individuals. The strict social distancing protocols were implemented in Norway at March 12. 2020 and the first data collection in this period (T1) has already been done. It lasted seven days and was between March 31st 2020 and April 7th 2020. Thus, the strict protocols had been held constant during the two weeks prior to data collection, as well as during the data collection week. Furthermore, no new information was given by the government during this period with regard to changes of distancing protocols, keeping expectation effects constant. From June 15., the majority of the distancing protocols will be lifted in Norway, and the T2 period is defined from this date. Data are set to be collected from the sample providing data in the first collection (N = 10 084), starting one week after the lifting of the protocols, that is, from 22th of June. The collection will last until enough data has been collected, but no longer than three weeks.

Ethical approval of the study was granted by The Regional Committee for Medical and Health Research Ethics and the Norwegian Centre for Research Data (reference numbers: 125510 and 802810, respectively, for the first data collection, and , respectively for the second data collection), where the study protocol and analysis plan was approved prior to data collection. The study is conducted in accordance with the guidelines of the Strengthening the Reporting of Observational Studies in Epidemiology statement (STROBE; Von Elm et al., 2007). The pre-registered protocol for a study of loneliness based on the first (T1) data collection can be found at Clinicaltrials.gov (Identifier: NCT04365881). The study is part of The Norwegian COVID-19, Mental Health and Adherence Project (Ebrahimi, Hoffart, \& Johnson, 2020).

Procedures The survey is disseminated online. At T1, it was disseminated in a systematic manner to give the adult population an equal opportunity to participate in the study. The dissemination procedure involved information about the survey through broadcasting on national, regional, and local news channels and provision of the online survey to a random selection of Norwegian adults on Facebook. The dissemination procedure is described in detail elsewhere (Ebrahimi et al., 2020).

The stopping rule for the first data collection was designed to ensure that the SDPs were held constant for two weeks prior to and the week during the data collection period, as well as controlling for expectation effects by stopping data collection instantly once information concerning forthcoming modification of the social distancing protocols were given. The second data collection - approaching the 10 084 individuals of the first collection - will start one week after the strict protocols are lifted (June 22.) and last three weeks.

Sample size and power The sample providing data at T1 comprised 10 084 participants, ascertaining power for the questions asked. All these participants will be invited to partake in the second data collection during the lifted prptocols period (T2) in accordance with the study plan. The mentioned 'Norwegian COVID-19, Mental Health and Adherence Project' involves multiple studies, where some involve a Complex Systems (Network analysis) approach. These mutlivariate analyses require large samples and power analysis was conducted accordingly. Following power analysis guidelines by Fried \& Cramer (2017), it is recommended that the number of participants be at the very least three times larger than the number of estimated parameters. However, more conservative recommendations by Roscoe (1975) for multivariate research, recommends sample size that is ten times larger than the number of estimated parameters. For our network of 9 PHQ-9 variables and 16 CAS-1 variables, the number of edges in the network is 25*24/2 = 300. Thus, according to the conservative advice of Roscoe, the required sample size is 3 000.

Blinding No blinding is involved in this study.

Is there any additional blinding in this study? Not applicable

Randomization No randomization.

Sampling Plan

Existing Data Registration prior to the analysis of the T1 data and prior to the collection of the T2 data.

Explanation of existing data The pandemics (and the consequent social distancing protocols) occurred, by its very nature, unexpectedly and the project could not be planned in advance. Thus, the first data collection started immediately and before pre-registration. It started during the time-period with the strictest protocols in Norway, and the first data collection was stopped before these protocols were modified or new information about them were added. We have avoided any network analysis of the data.

Manipulated variables Not applicable/NA

Statistical analysis To test H1, the network of the 9 PHQ items and the 16 CAS-1 items will be estimated using the R package qgraph (Epskamp et al., 2012) in a random half of the sample. Network stability tests will be conducted using R package bootnet (Epskamp, et al., 2018). An undirected partial correlation network will be estimated, resulting in associations between each pair of symptoms controlling for all other associations among symptoms. Centrality indices - strength and betweenness - will be computed according to the method of Robinaugh et al. (2016). Strength refers to how strongly a node relates to other nodes. Betweenness refers to how important a node is in paths between other nodes. Hypothesis 2 will be tested in the other random half of the sample. The outcomes of interest will be the degree to which change in a variable from T1 to T2 correlates with the change in the PHQ-9 total score (minus the variable examined if it is a symptom).

Missing data There are no missing data in the first data collection because the survey includes only mandatory responses. Dropout from the second data collection is to be expected, but the amount of missing data is uncertain. The multiple imputation procedure in SPSS will be used to impute missing data.

Exploratory analysis We may do other exploratory analysis that we have not yet thought of at the time of preregistration. If we do such exploratory analyses, we will explicitly state them as exploratory in the published manuscript following common publishing guidelines.

Other The data include one directly identifiable variable (contact information) for participants in accordance to the General Data Protection Regulation (GDPR) law in EU, which is to give the participants the opportunity to have their data deleted upon request. Data are thus kept on a safe server belonging to the University of Oslo and will be accessed first following de-identification.