Data from the control day (no intervention, no alcohol consumption on the previous day) of a study evaluating alcohol hangover effects were used for the current analysis [12]. The study was approved by the Psychology Ethics Committee of the University of Groningen (protocol code: ppo-015–002, approval date: 3 September 2015). Healthy volunteers (social drinkers) were recruited via local advertisement, and written informed consent was obtained prior to participation. Participants received €120,—for completing the study.

Inclusion criteria were being a healthy male or female, aged between 18 and 30 years. Exclusion criteria at screening included smoking, drug use, use of medication, or the presence of any underlying disease. On the test day, participants were further excluded if they experienced minor (immune-related) illness or reported poor sleep quality during the preceding night.

At the start of the test day, a medical examination was conducted by the study physician to confirm the participants’ health status and medical history. The use of illicit drugs (including amphetamines, barbiturates, cannabinoids, benzodiazepines, cocaine, and opiates) was verified via a urine drug test (AlfaScientic Designs Inc., Poway, CA, USA). Recent alcohol consumption was verified using the Alcotest 7410 Breath Alcoholmeter (Dräger, Hoogvliet, the Netherlands), ensuring a breath alcohol concentration of 0,00%.

The test day involved hourly assessments (09:30–15:30) of immune fitness and saliva collection for biomarker analysis. Between assessments, participants were instructed to relax (e.g., read a book). No food or drinks were allowed, except for a standardized breakfast at 09:00 and a standardized lunch at 12:00. Moderate, standardized amounts of water were provided if needed.

Immune fitness was assessed hourly (09:30—15:30) using a 10-cm visual analog scale (VAS), ranging from 0 (very poor) to 10 (excellent). Participants indicated their immune fitness score by placing and ‘X’ on the VAS, which was then measured in cm, with higher scores reflecting better immune fitness.

Saliva samples were collected hourly (09:30—15:30) using a passive drool method (SalivaBio’s Saliva Collection Aid, Salimetrics, State College, PA, USA). Samples were immediately stored at −80 degrees Celsius after collection.

Salivary concentrations (pg/ml) of IL-1β, IL-2, IL-4, IL-5, IL-6, IL-8, IL-10, granulocyte–macrophage colony-stimulating factor (GM-CSF), interferon-gamma (IFN-γ) and TNF-α were determined using a multiplex immunoassay (customized ProcartaPlex Immunoassay, ThermoFisher Scientific, Waltham, USA), following the manufacturer’s instructions and standard procedures described elsewhere [13].

For each multiplex plate, the lower limit of detection (LOD) was computed. For biomarker concentrations below the LOD, a value corresponding to half the LOD was used in the statistical analyses. Biomarkers with > 25% of assessments below LOD (i.e., IL-2, IL-4, IL-5, IL-10, GM-CSF, and IFN-α) were considered unreliable and excluded from further analyses.

All statistical analyses were conducted using SPSS (IBM Corp. Released 2013. IBM SPSS Statistics for Windows, Version 30. Armonk, NY, USA: IBM Corp.). The distribution of the data was evaluated via visual inspection and the Kolmogorov–Smirnov test and revealed that not all data was normally distributed. Immune fitness and biomarker concentrations (IL-1β, IL-6, IL-8, and TNF-α) assessed throughout the day (09:30–15:30) were compared to the morning assessment (09:30) using the Independent-Samples Kruskal–Wallis test. After applying Bonferroni’s correction for multiple comparisons, differences were considered significant if p < 0.0083.

Spearman’s correlations were computed to examine associations between immune fitness and biomarker concentrations at different time points. To account for the relatively small sample size, bootstrapping was applied (B = 10,000 samples), and bias-corrected and accelerated 95% confidence intervals (BCa 95% CIB) were calculated [14, 15]. The BCa 95% CIB ranges from − 1 to + 1, with narrower intervals indicating greater precision. Correlations were considered statistically significant if the BCa 95% CIB did not contain zero.