A total of 54 healthy females (30 HC users and 24 non-users; mean age = 25.7, SD = 4.4 years) were recruited through flyers posted at the University of Oslo, Oslo, Norway. Among current HC users, 17 (56.7%) used combined HCs, while 13 (43.3%) used progestin-only HCs.

Inclusion criteria included being aged between 18 and 40 years, possessing normal hearing, and currently not being pregnant or breastfeeding. Exclusion criteria included any previous or current psychiatric or neurological disorders, ongoing substance abuse, or use of psychoactive medications. Hearing was assessed through basic audiometric testing immediately before EEG recording.

All participants provided written informed consent before participation and received compensation equal to 30 US dollars for their participation. The study was approved by the Regional Ethics Committee, South-East Norway (REK ref.: 657156).

Participants were asked to indicate the reason for their use of HC, including both contraceptive and non-contraceptive indications. Among current users, 28 of 30 (93.3%) reported using HC solely to prevent pregnancy. Two participants (6.7%) reported use for menstrual pain relief (dysmenorrhea); one of these also reported a diagnosis of polycystic ovary syndrome (PCOS) and used HC as hormonal treatment for PCOS-related symptoms. No participants reported a history of endometriosis, premenstrual dysphoric disorder (PMDD), or amenorrhea. HC type (combined vs. progestin-only) was recorded, but the primary analysis pooled all current HC users to maximize power and estimate class-level effects. Participants using any hormonal contraceptive method (including oral pills, injectables, implants, and hormonal IUDs) were classified as current HC users. Age of HC initiation was recorded for both current and past users. Duration of current HC use was not collected, and cumulative exposure could therefore not be analyzed.

For participants reporting a natural menstrual cycle, cycle phase was determined by self-reported cycle day in a brief questionnaire. Phases were categorized as follicular, ovulatory, or luteal based on day ranges, consistent with prior questionnaire-based research.

The LDAEP was calculated from the N1, P2, and N1P2 peak-to-peak amplitudes at central midline electrode, Cz, based on the averaged AEP waveforms for each participant. Cz was chosen as the site for LDAEP calculation, as it has been commonly used in previous LDAEP studies (Hagenmuller et al., 2016; Hwang et al., 2021). The N1 amplitude was defined as the most negative peak occurring between 60 and 140 ms following stimulus onset, while the P2 amplitude was identified as the most positive peak within the 150–250 ms range. Peak identification was manually confirmed for all participants and conditions. For each component (N1, P2, and N1P2), LDAEP was quantified as the slope of a linear regression line fitted to amplitude values across the five stimulus intensities, reflecting the degree of amplitude modulation with increasing sound intensity. The N1P2 amplitude was calculated as the peak-to-peak difference between P2 and N1 at each intensity level, and its slope was derived from these values.

All analyses were conducted in R 4.5.0 (R Core Team, 2025) using base functions and the packages effsize (Torchiano, 2020), psych (Revelle, 2025), car (Fox and Weisberg, 2019), and emmeans (Lenth, 2025). Prior to inferential testing, variable distributions were assessed using the Kolmogorov–Smirnov test. LDAEP slope values (N1, P2, and N1P2) were approximately normally distributed within groups.

Unadjusted group differences in LDAEP slopes were evaluated using one-way analysis of variance (ANOVA). The primary analysis compared all current HC users (formulations pooled) with non-users, justified by the presence of exogenous hormones across methods. Exploratory contrasts between combined and progestin-only users were also conducted, though these subgroups were small (n = 17 vs. 13) and results interpreted cautiously. Although the comparison involved only two groups, ANOVA was used to align with the structure of the subsequent analysis of covariance (ANCOVA) models. Each slope was then reanalyzed using ANCOVA to adjust for age and depressive symptomatology (BDI-II somatic-affective and cognitive subscales). Analyses including depressive symptoms were pre-specified as secondary and exploratory, since participants were not recruited on the basis of mood symptoms. Differences in LDAEP slope values across menstrual cycle phases were analyzed using one-way ANOVA.

ANCOVAs were conducted using Type II sums of squares via the car package. Multicollinearity was assessed using variance inflation factors (VIFs), all of which were below 2.5. Adjusted group means were reported as estimated marginal means using the emmeans package. Effect sizes for both ANOVA and ANCOVA models were reported using partial η², with conventional benchmarks for interpretation: small = 0.01, medium = 0.06, large = 0.14.

Categorical variables (e.g., education level, contraceptive type, menstrual phase) were compared using Pearson’s chi-square test or Fisher’s exact test, depending on expected cell sizes. Age of HC onset and depressive symptom scores were analyzed using the Wilcoxon rank-sum test due to non-normal distributions. Subgroup comparisons (e.g., adverse mood effects, contraceptive type) were conducted using independent-samples t-tests or Wilcoxon tests, based on distributional assumptions. All tests were two-tailed, with alpha set to 0.05.

An a priori power analysis was conducted using G*Power 3.1 (Faul et al., 2009), based on the sample size and effect size (d = 0.77) reported in Andersson et al. (2025). The analysis indicated that a total sample size of 46 participants would be required to achieve 80% power at an alpha level of 0.05.

Table 1 presents descriptive statistics for age, education level, menstrual phase, age of HC usage onset, and self-reported adverse mood effects across the three initial groups: current HC users (n = 30), past users (n = 17), and never-users (n = 7). The groups did not significantly differ in age, F(2, 51) = 2.14, p = 0.129, or in education level, χ²(4) = 8.01, p = 0.091. Among participants who reported having a menstrual cycle and were aware of their cycle phase (n = 40), no significant group differences were found in the distribution across follicular, ovulatory, and luteal phases, χ²(4) = 1.28, p = 0.865. Likewise, the age of HC initiation did not significantly differ between current and past users (W = 280.5, p = 0.572, Wilcoxon rank-sum test). Duration of current HC use was not available. Three participants reported non-contraceptive indications for HC use: two for dysmenorrhea and one for PCOS. Single-case tests showed that their LDAEP slopes did not differ significantly from other HC users (all p ≥ 0.05). Sensitivity ANCOVAs excluding these participants confirmed that the main HC effect remained significant.

A key objective of this study is to isolate the neurophysiological effects of ongoing HC usage. To our knowledge, there are no longitudinal studies showing whether central serotonergic markers normalize after discontinuation. Available evidence is limited to cross-sectional comparisons of current users and non-users. In this sample, all but two former HC users had ceased usage for at least 1 year. We therefore assessed comparability empirically in this dataset.

Initial analyses also indicated that past users did not differ significantly from never-users in key demographic variables (e.g., age, education) or clinical measures (including LDAEP and depressive symptoms). Omnibus ANOVAs confirmed group effects on N1 and N1P2 slopes, and post hoc (Tukey) comparisons showed steeper slopes in current users versus past users, while past and never users did not differ. Moreover, the never-user subgroup was notably small (n = 7), diminishing the statistical power and interpretive clarity for comparisons among the three groups. By merging past users and never-users into a single “non-users” group, the study achieves both a more robust sample size and a clearer contrast against current HC users. This approach enhances the statistical power to detect meaningful differences attributable to active HC usage and focuses the analysis on the immediate effects of exogenous hormones on central serotonergic activity.

Figure 1 displays the grand-average ERP waveforms at Cz for current HC users and non-users in response to each of the five stimulus intensities. LDAEP slopes (N1, P2, and N1P2), measured at Cz, differed significantly between current HC users and non-users, as illustrated in Figure 2.

For the N1 LDAEP slope, current HC users exhibited a significantly more pronounced negative slope (M = −0.080, SD = 0.033) compared to non-users (M = −0.052, SD = 0.040). This difference was statistically significant, F(1, 52) = 8.05, p = 0.006, with an associated partial η² of 0.13 [95% CI (0.02, 1.00)]. The mean difference between groups was −0.028 [95% CI (−0.048, −0.008)].

For the P2 LDAEP slope, current HC users exhibited a steeper LDAEP slope (M = 0.120, SD = 0.068) than non-users (M = 0.084, SD = 0.046). This difference was statistically significant, F(1, 52) = 5.15, p = 0.028, with an associated partial η² of 0.09 [95% CI (0.01, 1.00)]. The mean difference between the groups was 0.036 [95% CI (0.004, 0.069)].

For the N1P2 LDAEP slope, current HC users exhibited a steeper slope (M = 0.200, SD = 0.084) compared to non-users (M = 0.136, SD = 0.064). This difference was statistically significant, F(1, 52) = 9.59, p = 0.003, with an associated partial η² = 0.16 [0.03, 1.00]. The mean difference between the groups was 0.064 [95% CI (0.023, 0.106)].

After covariate adjustment, LDAEP slopes measured at Cz remained higher in current HC users than in non-users.

N1 LDAEP slope differed significantly between current HC users and non-users after controlling for age and depressive symptom scores, F(1, 49) = 7.68, p = 0.008, partial η² = 0.14. Neither covariate was statistically significant. Adjusted estimated marginal means were −0.080 (SE = 0.007) for users and −0.052 (SE = 0.007) for non-users. The regression model yielded R² = 0.209, adjusted R² = 0.144.

P2 LDAEP slope differed significantly between current HC users and non-users after controlling for age and depressive symptom scores, F(1, 49) = 5.87, p = 0.019, partial η² = 0.11. Age and depressive symptoms were not significant predictors. Adjusted estimated marginal means were 0.122 (SE = 0.011) for users and 0.081 (SE = 0.012) for non-users. The regression model yielded R² = 0.116, adjusted R² = 0.044.

N1P2 LDAEP slope was significantly higher in current HC users than in non-users after controlling for age and depressive symptom scores, F(1, 49) = 10.13, p = 0.003, partial η² = 0.17. Neither covariate showed a significant association. Adjusted estimated marginal means were 0.202 (SE = 0.014) for users and 0.133 (SE = 0.016) for non-users. The regression model yielded R² = 0.199, adjusted R² = 0.134.

As a secondary, exploratory aim, we examined whether current HC users differed from non-users in self-reported depressive symptoms by comparing BDI-II scores between the groups. Wilcoxon rank-sum tests showed no significant differences in total BDI-II, cognitive, or somatic-affective subscale scores (all p > 0.264).

To assess whether mild depressive symptoms modulated LDAEP, participants were stratified by a validated BDI-II threshold into subthreshold (n = 33) and above-threshold (n = 21) groups. Among current HC users, 17 were subthreshold and 13 above-threshold; among non-users, 16 were subthreshold and 8 above-threshold. Independent-samples t-tests comparing LDAEP slopes (N1, P2, and N1P2) between sub- and above-threshold participants across the full sample revealed no significant differences (all p > 0.209). Additional within-group comparisons, conducted separately for HC users and non-users, also showed no significant differences in LDAEP slopes between subthreshold and above-threshold individuals.

Further exploratory subgroup analyses were conducted to examine whether menstrual phase or HC type was associated with LDAEP modulation. Among non-users, one-way ANOVA across self-reported menstrual cycle phases (follicular, ovulatory, and luteal) revealed no significant differences in N1, P2, or N1P2 slopes (all p > 0.170). Equivalent results were found in current HC users. Given that cycle phase was assessed by self-report without hormonal verification, these analyses should be considered exploratory. Additionally, among current HC users, no significant differences were observed between those using combined HCs (n = 17) and those using progestin-only HCs (n = 13) across any LDAEP component. Independent t-tests showed no significant group differences in N1, P2, or N1P2 slopes.

Adverse mood effects attributed to HC use were reported by 11 out of 30 current HC users (36.7%) and by 14 out of 17 past users (82.4%) during their previous use. Although LDAEP slopes (N1, P2, N1P2) did not differ significantly between those with and without such reported side effects (all p ≥ 0.235), exploratory analyses of depressive symptoms suggested group differences. Specifically, participants who endorsed adverse mood effects scored significantly higher on total BDI-II scores (W = 158.5, p = 0.021) and on the somatic-affective subscale (W = 172, p = 0.004), whereas cognitive scores did not differ (W = 134.5, p = 0.201). Furthermore, a Fisher’s exact test revealed that adverse mood effects were significantly associated with being above the clinical BDI-II threshold (p = 0.040). Given the exploratory nature of these subgroup analyses, findings should be interpreted with caution.

The findings of the present study highlight the importance of accounting for HC use in future research involving LDAEP. Given that HC users showed consistently elevated LDAEP slopes, indicating reduced central serotonergic activity, ignoring HC status may obscure other true group differences or introduce systematic bias. Moreover, the absence of a significant association between LDAEP and depressive symptoms in this healthy, non-clinical sample highlights the need to investigate these dynamics in clinical populations. Specifically, future studies should explore whether HC-related serotonergic alterations contribute to depressive symptomatology in vulnerable individuals or whether a subset of HC-using women with depression may exhibit a distinct serotonergic profile identifiable via LDAEP. Such work could help clarify whether HC use not only modulates central serotonergic activity but also plays a pathophysiological role in a specific subtype of depression. This line of research may ultimately enhance the precision of LDAEP as a diagnostic or predictive tool and improve the understanding of how synthetic sex hormones interact with neurobiological pathways relevant to mood disorders.

Although this study did not investigate SSRI treatment outcomes directly, the elevated LDAEP values observed among current HC users suggest central serotonergic downregulation that may have implications for antidepressant response. High LDAEP has consistently been associated with better SSRI outcomes in depressed patients (Ip et al., 2023; Yoon et al., 2021), presumably because SSRIs can effectively augment central serotonergic tone when it is initially low. However, findings from Larsen et al. (2022) complicate this relationship by showing that HC use moderates the link between central serotonergic function and SSRI efficacy. In their sample of depressed women, lower 5-HT₄ receptor binding, an index of reduced serotonergic activity, was associated with better treatment response in non-users, but with worse response in HC users. This crossover interaction implies that HC users may represent a distinct serotonergic-related subtype of depression with qualitatively different treatment trajectories. Our findings support this framework by showing that central serotonergic function, as indexed by LDAEP, is already altered in healthy HC users, suggesting that the serotonergic modulation associated with HC use may precede clinical depression. Taken together, these results underscore the need to consider HC status when interpreting central serotonergic markers like LDAEP, particularly in relation to SSRI treatment response.

A key strength of this study is the demographic and clinical homogeneity of the groups, which increases internal validity by minimizing confounds. Data quality was also high, with minimal artefact rejection and use of standard EEG preprocessing pipelines. Conversely, the modest sample size may provide limited statistical power to detect marginal effects. Our primary estimand was the class-level effect of HC use, but subgroup sizes for combined vs. progestin-only users (n = 17 vs. 13) were limited, and exploratory contrasts are interpreted cautiously. The presence of one participant with PCOS and two with dysmenorrhea is another limitation, though sensitivity analyses confirmed that excluding them did not change the results. Reliance on self-reported measures (BDI-II, retrospective mood effects, menstrual cycle phase) may introduce bias and reduces precision, particularly for cycle-phase comparisons. We also lacked data on duration of current HC use; age of initiation was recorded but is only a rough proxy for cumulative exposure. Finally, heterogeneity in LDAEP acquisition and processing protocols across studies complicates direct comparisons; greater methodological standardization would aid reproducibility.