Nervous System, Autonomic

The following first-order factor models were tested: the Razavi [31 (link)] model with a single one order factor; the Zigmond-Snaith [1 (link)] model with odds and even items for anxiety and depression, respectively; the Moorey [21 (link)] model with Anxiety (Items 1, 3, 5, 9, 11, and 13) and Depression (Items 2, 4, 6, 7, 8, 10, 12, and 14); the Dunbar [7 (link)] model with Anhedonic Depression (Items 2, 4, 6, 8, 10, 12, and 14), Autonomic Anxiety (Items 3, 9, and 13), and Negative Affectivity (Items 1, 5, 7, and 11); the Friedman [10 (link)] model with Depression (Items 2, 4, 6, 8, 10, 12, and 14), Psychic Anxiety (Items 3, 5, 9, and 13) and Psychomotor Agitation (Items 1, 7, and 11); the Caci [13 (link)] model with Depression (Items 2, 4, 6, 8, 10, 12, and 14), Anxiety (Items 1, 3, 5, 9, and 13) and Restlessness (Items 7, 11, and 14). In addition, we tested a bifactor model, with a general factor and two group factors with Anxiety (Items 1, 3, 5, 7, 9, 11, and 13) and Depression (Items 2, 4, 6, 8, 10, 12, and 14).
Since the data were not normally distributed (Mardia’s normalized coefficient = 41.68), maximum likelihood (ML) robust estimators were used. Accordingly, we reported fit statistics based on the Satorra–Bentler scaled chi square (SBχ²) as available in EQS 6.2 [32 ]. Because of the large sample size, we expected all models to have a significant chi-square value. Therefore, more “practical” indices of fit were used to evaluate each model’s fit as well as to compare alternative models, according to the recommended cut-offs [33 ,34 ]. More specifically, a chi-square to degree of freedom ratio value is used to minimise the impact of sample size on the model chi-square; values less than 2 indicate good fit. The Akaike Information Criterion (AIC) is a statistic generally used to compare the fit of non-nested or non-hierarchical models; lower values indicate a better fitting model. Both the comparative fit index (CFI) and the non-normed fit index (NNFI) result from a comparison between the hypothesized model’s chi square with the independence model’s one. Values greater than .95 are recommended for both indices. The root mean squared error of approximation (RMSEA) is instead a ‘badness of fit’ index assessing the difference between the reproduced covariance matrix and the population covariance matrix. RMSEA very close to 0 indicate almost perfect fit; values less than .05 are recommended as they reflect a small approximation error. The 90% confidence interval (CI) around the RMSEA point estimate is also commonly reported to indicate the possibility of close or exact fit.
Some physical and mental health outcomes (i.e. depression) show a different trend according to variation in age and gender, namely older adults frequently score higher in depression [35 (link)]. Therefore, we tested the invariance of the best-fitting model across different gender and age groups. A first multi-group analysis was based on two groups comprised of 770 males and 829 females, respectively. Then, a second multigroup analysis was based on two age groups composed of 685 and 914 participants aged under and over 45 years, respectively.

In subcohort I, we will record resting state EEG and event-related potentials (ERPs) with simultaneous two-lead electrocardiography (ECG) to measure autonomic nervous system activation. EEG will be recorded using a 256-channel HydroCel Sensor Net system (MagstimEGI, USA) at 1000 Hz, where the vertex electrode serves as the reference. Impedances across all electrodes will be kept below 50 kΩ. ECG will be acquired at 1000 Hz using a Physio 16 device (MagstimEGI, USA). EEG/ERP recording: resting EEG (6 min eyes closed and eyes open), two-tone auditory oddball and the LDAEP tasks.

HRV, the fluctuation of heartbeat intervals measured using an electrocardiogram, is used to evaluate autonomic nerve activities [26 (link), 27 (link)]. HRV tends to be lower in a person with anxiety or depression. However, it is relative rather than absolute; therefore, it is not directly compared among individuals.
Standard deviation of the normal- to- normal interval (SDNN) is the quantification of HRV to further compare it among individuals. Particularly, SDNN is the standard deviation of the R-R intervals of the heartbeat in a certain time duration and is obtained via time-domain analysis. SDNN was used to evaluate cardiovascular compatibility. SDNN includes all the different types of variations and represents total variability [28 (link)]. It assesses the flexibility of the autonomic nervous system and the balance of sympathetic and parasympathetic nervous systems, with an increase in SDNN reflecting the stability of these systems [16 (link), 29 (link)]. The grand mean of SDNN scores among resting adults is 50 mseconds [29 (link)].

Data are presented as means ± SEM and submitted to Shapiro-Wilk homogeneity variance test. Treadmill performance and body weight in S and T SHR and Wistar rats were analyzed by three-way ANOVA with repeated measurements (time). Differences in hemodynamic and autonomic parameters, BBB permeability, ultrastructural changes in BBB constituents, and caveolin-1 expression between groups and conditions were analyzed by two-way factorial ANOVA. Tukey was the post hoc test. Correlations analyses used the Pearson statistics. All statistical analyses were performed by the GraphPad Prism 8 software. Differences were considered significant at p < 0.05.