Study design
The current study utilised cross-sectional baseline data from the Age-Well randomised clinical trial of the Medit-Ageing European project. Details of the recruitment method and eligibility criteria can be found in the Age-well protocol35.
Participants
137 participants were recruited from the general population (Caen, France), screened, and included in the trial. All participants were required to be aged 65 or over, have at least seven years of education, perform normally on standardised cognitive tests according to study-specific standards (age, sex, and education level) and be retired for at least one year. Key exclusion criteria included the presence of major neurological or psychiatric disorder, chronic disease or acute unstable illness that may interfere with cognitive functioning, history of cerebral disease, and current or recent medication that may interfere with cognitive function. Further details relating to inclusion and exclusion criteria can be found in the trial protocol paper35.
Written informed consent was obtained from all participants. The Age-Well randomized clinical trial was sponsored by the Institut National de la Santé et de la Recherche Médicale (Inserm), approved by the ethics committee (Comité de Protection des Personnes Nord-Ouest III, Caen, France; trial registration number: EudraCT: 2016-002441-36; IDRCB: 2016-A01767-44) and registered on ClinicalTrials.gov (Identifier: CT02977819).
Repetitive negative thinking (RNT)
Worry. Worry was measured using the self-report 16-item Penn State Worry Questionnaire (PSWQ)36. Each item is rated on a 5-point Likert scale, ranging from 1 (“not at all typical of me”) to 5 (“very typical of me”). Total scores range from 16 to 80, with higher scores being indicative of higher levels of worry. Previous research has demonstrated good internal consistency and adequate convergent validity in older adults37.
Ruminative brooding. Ruminative brooding was measured using the 5-item brooding subscale of the self-report 22-item Ruminative Response Scale (RRS)38. Each item is rated on a 4-point Likert scale, ranging from 1 (“almost never”) to 4 (“almost always”). Total scores range from 5 to 20, with higher scores being indicative of higher levels of ruminative brooding. Previous research has demonstrated acceptable internal consistency in older adults39.
Neurofilament Light
Non-fasting blood samples were collected from participants between 7.30am and 8am. Samples were drawn from the antecubital vein into a 5ml EDTA bottle and then, after decantation and collection of plasma, stored in a -80°C freezer before analysis. Plasma levels of NfL were performed by an ultra-sensitive electrochemiluminescence measurement technic (Meso Scale Discovery, MSD, Rockville, USA) using R-PLEX Human Neurofilament L Antibody Set with reference ranges of 5.5 to 50,000 pg/mL. All analyses took place in the Centre de Ressources Biologiques (CRB) in Caen, France.
Allostatic load
The allostatic load composite created in the current study was comprised of five sub-categories: anthropometric, cardiovascular & respiratory, metabolic, immune, and neuroendocrine. Detailed descriptions of the measures included in the composite are provided in the Supplementary Materials. Briefly:
Anthropometric category included measures of body mass index (BMI) and waist-hip ratio (WHR).
Cardiovascular and respiratory category included measures of systolic and diastolic blood pressure (SBP, DBP), pulse pressure, standard deviation of the average heartbeat-to-beat intervals (SDANN) and root mean square of successive differences between normal heartbeats (RMSSD).
Metabolic category included measures of plasma insulin, serum triglycerides, high and low-density cholesterol (HDL, LDL) and serum creatinine.
Immune category included plasma measures of interleukine-6 (IL-6) and C-reactive protein (CRP).
Neuroendocrine category included serum cortisol, dehydroepiandrosterone sulfate (DHEA-S), plasma norepinephrine (NE) and epinephrine (E).
Measures within each sub-category were z-transformed and averaged to create sub-category scores. The average of each of the five sub-categories was then calculated and re-standardised to form the total allostatic load score, resulting in a mean of 0 and a standard deviation of 1. Values for HDL, DHEA-S, SDANN, and RMSSD were reverse-scored before being included in the composites, so that higher scores indicated higher levels of allostatic load. Participants with missing scores on any of the measures were not included in the composite score calculation.
Covariates for sensitivity analyses
To examine the specific effects of the cognitive component of RNT (i.e., ruminative brooding and worry), we conducted sensitivity analyses controlling for depression (for ruminative brooding) and anxiety (for worry) due to their close association with RNT40,41.
Depression. Depression was measured using the self-report 15-item Geriatric Depression Scale (GDS-15)42. Each item is answered with either “yes” or “no”. Total scores (calculated by summing the number of “yes” responses) range from 0 to 15, with higher scores being indicative of higher levels of depression. Previous research has demonstrated good internal consistency43 and validity42 in older adults.
Anxiety. Anxiety was measured using the 20-item trait anxiety subscale of the State-Trait Anxiety Inventory (STAI-B)44. Each item is rated on a 4-point Likert scale, from 1 (“almost never”) to 4 (“almost always”). Total score for STAI-B subscale (calculated by summing all the respective item scores) range from 20 to 80, with higher scores being indicative of higher levels of anxiety. Previous research has demonstrated good internal consistency and validity for older adults45.
Statistical analyses
From the 137 participants enrolled in Age-Well, two participants were excluded from all secondary analyses for not meeting major eligibility criteria as determined by the Trial Steering Committee (one participant had experienced a head trauma with loss of consciousness for more than one hour and one participant developed amyotrophic lateral sclerosis over the course of the trial). One participant had missing data for NfL. This resulted in 134 participants being included in the analyses with NfL. Participants with missing data for at least one of the measures included in the allostatic load composite were excluded in analyses with allostatic load.
Associations between RNT, NfL, allostatic load and potential confounds (i.e., demographic characteristics and psychiatric symptoms) were investigated using Pearson’s correlations for continuous variables, and t-tests for sex differences.
Separate linear regression analyses were run for each RNT measure (i.e., worry and ruminative brooding) to examine their relationships with NfL. Model 1 was unadjusted, Model 2 adjusted for age and sex, and Model 3 examined the interaction with sex, adjusting for age. Sensitivity analyses were run to determine whether any associations remained after adding anxiety to Model 2 and Model 3 when worry was the predictor or depression when ruminative brooding was the predictor. Given the low levels of depressive symptoms observed in the sample, analyses were conducted using depression as a categorical variable (participants with no depressive symptoms [i.e., scores of 0] and participants with any depressive symptoms [i.e., scores of 1 or above])46.
Finally, to explore the potential mechanism linking RNT to NfL, we examined the relationship between worry and ruminative brooding and allostatic load in unadjusted and adjusted regressions (i.e., Model 1 and Model 2, as above); education was added as a covariate in Model 2 because allostatic load has previously been associated with education47. Model 3 examined the interaction with sex, adjusting for age and education. In the case of a significant association being observed between worry and/or brooding and NfL, and between worry and/or brooding and allostatic load, a mediation analysis would be conducted to investigate whether allostatic load mediated the relationship between association between worry and/or brooding and NfL.
Exploratory analyses were conducted to examine the associations between RNT and each sub-domain of the allostatic load composite.
All analyses were performed with in R (version 4.2.2). Standardised beta coefficients are reported.