Open-label placebo effects on psychological distress and hair cortisol in a randomized controlled trial – the moderating role of the five-factor personality traits

Abstract

Objectives

Open-label placebo (OLP) treatments have shown promising effects on various health outcomes, but evidence regarding physiological measures is still limited. Personality traits have been proposed as important moderators of treatment responsiveness, yet their role in OLP effects remains largely unclear. This study therefore investigated whether the five-factor personality traits moderate the effect of a four-week OLP intervention to reduce hair cortisol concentrations (HCC) and psychological distress in the context of oral university exams as real-life stressor.

Methods

In a randomized controlled trial, 202 healthy university students were assigned to either a four-week OLP treatment or a no-treatment control group. Psychological distress was assessed repeatedly across the intervention period, while HCC was measured pre- and within the intervention. Moderated regression analyses examined the influence of personality traits on OLP responsiveness, with conscientiousness and neuroticism as hypothesized moderators.

Results

Compared to the control group, the OLP group reported significantly lower psychological distress and showed a significant decrease in HCC. Conscientiousness moderated group effects on psychological distress, with higher trait levels predicting stronger reductions in the OLP group. Neuroticism moderated group effects on HCC, with highly neurotic individuals in the OLP group showing the strongest decreases. Exploratory analyses indicated no significant moderation effects of extraversion, agreeableness, or openness.

Conclusion

These findings demonstrate that OLPs can reduce both subjective and physiological distress measures in a naturalistic stress context. Personality traits play a differential moderating role across subjective and physiological measures, suggesting that OLPs may be particularly beneficial for vulnerable individuals.

Highlights

• •Open-label placebos reduce both psychological distress and hair cortisol.
• •Conscientiousness moderates the open-label placebo effect on psychological distress.
• •Neuroticism moderates the open-label placebo effect on hair cortisol concentrations.
• •Extraversion, agreeableness, and openness show no moderating effects.
• •Stress-vulnerable individuals benefit most from open-label placebos.

1. Introduction

Placebo treatments are traditionally employed as control conditions in clinical research to determine the efficacy of pharmacological or behavioral interventions. Apart from that, placebos can elicit clinically relevant effects, including reductions in pain, stress, and anxiety [1,2], and are frequently used in routine clinical practice [3]. However, the conventional deceptive placebo administration, where patients expect to receive a medical treatment, has raised ethical concerns about informed consent and patients' autonomy [4].

To address these concerns, recent research has focused on open-label placebo (OLP) interventions, in which individuals are informed that they receive a placebo. OLPs have shown beneficial effects across various self-reported outcomes, including irritable bowel syndrome (IBS) symptoms [5], back pain [6], and psychological distress [[7], [8], [9]], with meta-analyses reporting small-to-medium-sized effects [10,11]. Furthermore, OLPs may achieve effects comparable to deceptive placebos [12,13], challenging the assumption that deception is necessary for placebo effects.

However, studies examining OLP effects on physiological measures, such as stress-related biomarkers, remain limited and inconclusive [[14], [15], [16]]. Several considerations highlight the importance of addressing this gap. First, given that studies have demonstrated deceptive placebo effects on various physiological indicators [17,18], meta-analyses underscore the urgent demand for OLP research including physiological outcomes to examine the validity of OLP effects beyond subjective reports [10,19]. Second, subjective and physiological stress measures often show weak or inconsistent associations [20], suggesting that self-reports do not fully capture all components of psychophysiological stress reactivity. Stress responses engage multiple systems, including the hypothalamic–pituitary–adrenal (HPA) axis as the primary neuroendocrine stress system [21], with cortisol as established marker measurable in saliva to capture acute stress responses [22], while hair cortisol concentrations (HCC) reflect long-term or chronic stress [23]. Third, naturalistic stress contexts involving performance demands and social-evaluative threat (e.g., university exams) are particularly relevant for assessing physiological responses, as they typically elicit robust HPA-axis activation [22]. However, OLP studies have not yet investigated physiological measures in such real-life settings [[7], [8], [9]], highlighting the lack of evidence regarding whether OLPs can modulate psychophysiological stress responses under naturalistic conditions.

Findings from conventional placebo research indicate that personality traits are associated with variability in placebo responsiveness [17,18,24]. Personality traits predict a wide range of life and health-related outcomes, including treatment responsiveness [25]. The Five-Factor Model (FFM) provides a worldwide accepted, empirically validated, comprehensive taxonomy of personality, comprising the dimensions extraversion, agreeableness, conscientiousness, neuroticism, and openness to experience [26]. Evidence suggests that personality may modulate placebo responses depending on contextual factors [27,28]. For example, neuroticism is associated with reduced placebo responses in experimental pain [17], unrelated to placebo effects in IBS [24], and positively linked to placebo responsiveness in psychosocial stress [27], highlighting the contextual variability. Therefore, personality traits should be examined under consideration of specific contexts rather than assuming uniform associations across placebo effects.

Whether the FFM traits similarly influence OLP responses has been examined in only two studies so far. Ballou et al. [29] reported an association between agreeableness and OLP responses in IBS when controlling for IBS-specific predictors, whereas extraversion and openness were unrelated to OLP responses. In experimental heat pain, openness was not significantly associated with OLP responses [30]. To our knowledge, the influence of conscientiousness and neuroticism has not yet been investigated in OLP research.

In academic stress contexts, conscientiousness and neuroticism may represent potential moderators of OLP responses. Conscientiousness reflects self-control, responsibility, diligence, orderliness, and goal-directed behavior [31], and is associated with better academic performance, self-efficacy, and achievement motivation [32,33]. Although conscientiousness generally promotes stress resilience [34], social-evaluative or performance-related stressors may increase emotional tension and stress among highly conscientious individuals [35,36], potentially depending on the degree of controllability during stress exposure. In treatment contexts, conscientiousness is positively linked to medication adherence [37] and favorable treatment outcomes [25], suggesting enhanced engagement with and responsiveness to interventions, including OLPs.

Neuroticism, characterized by emotional volatility, anxiety, and negative affectivity [31], is associated with poorer academic performance [38], higher psychological distress [39], and heightened physiological stress reactivity, reflected in prolonged sympathetic activation [40], greater cardiovascular responses [41], and elevated salivary and hair cortisol [42,43]. In academic stress contexts, neuroticism is linked to increased anxiety [44] and burnout [45], underscoring the emotional salience of evaluative situations for highly neurotic individuals. While neuroticism is associated with less favorable treatment outcomes in clinical samples [25], non-clinical deceptive placebo studies suggest a more nuanced pattern [27,46], with stronger placebo responses reported among highly neurotic individuals during psychosocial stress. It is suggested that highly neurotic individuals may benefit more from certain interventions, possibly because their elevated distress allows more space for improvements or enhance their treatment engagement [25]. Consequently, highly neurotic individuals may be particularly responsive to OLPs in stress contexts, especially when the placebo treatment directly targets their core concern: anxiety and stress regulation.

With regard to the remaining FFM traits, individuals low in extraversion show higher stress vulnerability and may therefore benefit more from OLP-induced improvements in academic distress [34]. Individuals high in openness may be more inclined to engage with alternative coping strategies, whereas individuals high in agreeableness may exhibit greater compliance and cooperativeness, both of which could enhance treatment responsiveness [25]. However, these assumptions require exploratory, empirical examinations due to the currently insufficient accumulated evidence [29,30].

In sum, neuroticism and conscientiousness not only influence how individuals respond to (academic) stressors but also modulate treatment responsiveness. Investigating their moderating roles in OLP responses under real-life stress conditions may help identify for whom such interventions are more beneficial and contribute to explaining the psychological mechanisms underlying OLP effects. Furthermore, investigating both subjective and physiological stress measures may provide a comprehensive, multi-level perspective on OLP effects. Thus, the present study aims to address four objectives: (1) to investigate whether a four-week OLP intervention during an university exam period reduces psychological distress and HCC compared to a control group, (2) to test whether higher conscientiousness and (3) higher neuroticism are associated with stronger OLP effects, and (4) to examine the moderating roles of extraversion, agreeableness, and openness on OLP responses in exploratory analyses.

2. Methods

2.1. Sample

The final sample comprised 202 healthy university students (mean age: 26.03 ± 4.53 years; 157 females [77.70%]; see Fig. 1 for CONSORT flowchart) who were randomly assigned to the OLP (n = 100) or the control group (n = 102). Participants were recruited via mailing lists and announcements in lectures. Eligibility criteria were an upcoming oral exam in the participant's respective field of study (health-related study programs) and a minimum hair length of 3 cm. Exclusion criteria were (i) factors related to the recruitment of a healthy sample: current or past psychiatric, neurological, or endocrinological disorders or treatments, diabetes, and regular consumption of illegal drugs; (ii) factors relevant to cortisol assessment: heavy smoking (>10 cigarettes/day), corticosteroid medication use, and pregnancy; and (iii) factors related to the OLP administration: allergies to sugar, artificial additives, or colorings. Due to that, n = 3 participants with ongoing psychiatric treatment were excluded from the analyses. In accordance with clinical trials ≥80% of pills taken defines good adherence [47]. Participants with lower adherence were excluded from the analyses (n = 3). After the trial completion, all participants received monetary compensation of 40 euros and six course credit hours.

Fig. 1.

CONSORT flowchart of participants. T5 denotes the assessment on the day of the oral exam, indicating that participants attended the exam.

2.2. Experimental design

In this randomized controlled trial (RCT), healthy university students were allocated to either an OLP intervention or a no-treatment control group (e.g., Refs. [7,8]). Prior to participation, all individuals provided written informed consent. The study received ethical approval from the MSB Medical School Berlin ethics committee (dossier MSB-2021/60), and was conducted in accordance with the Declaration of Helsinki. It was funded by the Deutsche Forschungsgemeinschaft (DFG, German Research Foundation) and registered in the German Clinical Trials Register (DRKS00031423), encompassing a research project with several planned publications addressing different research questions. A previous publication examined psychophysiological OLP effects in an initial subsample of the same longitudinal RCT [48]. The present article extends this work by using the expanded final sample, which includes all assessment waves, to examine the FFM personality traits as moderators of psychophysiological OLP effects. The registration does not include a pre-specified analysis plan; however, the analyses were guided by the predefined analyses of the DFG-funded research project. Given the limited evidence for OLP effects on biomarkers, we based the estimated effect size for an a priori power analysis on the smallest meta-analytic OLP effect [10], corresponding to f ≈ .215. Using this estimate and assuming an α error probability of .05, a desired power of .8, the required sample size of N = 172 was calculated.

2.3. Study procedure

Baseline measures of psychological distress, personality traits, and demographic characteristics were assessed in the university laboratory five weeks before the oral exam and prior to randomization (T0). Subsequently, all participants received written instruction about the OLP rationale analogue to previous OLP trials [5,6]. This instruction stated that (1) placebos can be powerful, (2) a possible mechanism may be classical conditioning, (3) a positive attitude is advantageous but not necessary, (4) a conscientious and regular intake is important for the effectiveness. The participants were further informed that non-deceptive placebos may have positive effects and watched a video report on OLPs (similar to Ref. [6]). Then the first hair sample was collected.

Next, randomization was conducted using sealed, opaque envelopes that contain the assignment to either the OLP or control condition. To ensure allocation concealment, student research assistants, who were blind to the envelope contents, administered the randomization. Each participant selected one envelope, resulting in a 1:1 random assignment to the respective condition. Participants were informed about their group assignment, while researchers and principal investigators remained blinded.

Participants assigned to the OLP condition received a package with 56 round placebo pills with a diameter of 7 mm (“P-pills" white, Lichtenstein manufactured by Zentiva Pharma GmbH), labeled with the university logo and the instruction “Placebo-pills, take two daily, in the morning and evening, over four weeks”. Participants were instructed to take the first placebo pill 28 days before the exam in the evening and the last pill on exam day in the morning. To ensure equal interactions and contact frequency across groups, weekly reminders regarding pill administration for OLP group were included in the routine online questionnaires sent to both, the OLP and the control group. Participants assigned to the control group did not receive any intervention but were informed about the option to obtain placebo pills after the trial, however, none opted to receive them.

Over the course of the subsequent four weeks, psychological distress was assessed at regular intervals for both groups via online questionnaires, specifically 28 (T1), 21 (T2), 14 (T3), and 7 days (T4) before the exam. On the day of the exam (T5), participants completed self-report measures of psychological distress in the university laboratory immediately before and after the 20-min oral exam. Participants in the OLP group were asked to return their leftover pills. Lastly, the second hair sample was collected.

2.4. Outcome measures

Psychological distress was conceptualized using the self-report measures of negative affect, test anxiety, subjective stress and exam-related stress. The 10-item subscale negative affect (NA) of the German Positive and Negative Affect Schedule (PANAS) was used to assess negative affect, referencing the past week during the intervention period [48]. Test anxiety was evaluated with the 20-item German version of the Test Anxiety Inventory (PAF), assessing thoughts and emotions with respect to exams [49]. To measure general subjective stress over the past week, we used the 7-item stress subscale from the German Depression Anxiety and Stress Scale 21 (DASS-21; [50]). In addition, participants rated their perceived exam-related stress using a visual analogue scale (VAS; [51]), indicating how stressful they perceive the current situation in the context of the upcoming exam (range 0-100). The five-factor personality traits were assessed using the German version of the Big Five Inventory 2 (BFI-2), which contains the dimensions extraversion, agreeableness, conscientiousness, neuroticism, and openness to experience [52]. The questionnaire consists of 60 items (12 per dimension), which are answered on a five-point Likert scale (1 ‘strongly disagree’ to 5 ‘strongly agree’).

To determine HCC, hair samples were collected at two time points: 35 days before the exam (T0) and on the exam day (T5). Two fine strands were taken from the posterior vertex using a loop, cut close to the scalp, and stored in aluminum foil [53]. As human hair grows approximately 1 cm per month [54], the 1 cm segment nearest to the scalp was analyzed. Accordingly, the T0 segment reflects pre-intervention cortisol levels, while the T5 segment represents cortisol levels during the intervention month. Sample preparation and cortisol extraction were performed by an independent laboratory (Dresden LabService GmbH, Dresden, Germany) following the protocol of Gao et al. [53], using liquid chromatography coupled with tandem mass spectrometry (LC-MS/MS), the primary method for steroid hormone quantification. Samples were analyzed in five batches corresponding to the five data collection semesters, with an average storage time of 6.65 weeks (SD = 3.19). HCC was not significantly associated with batch (HCC T0: r = −.036, p = .615; HCC T5: r = −.013, p = .854) and storage time (HCC T0: r = .019, p = .787; HCC T5: r = −.003, p = .964). The intra-assay variance coefficients were between 3.8 and 12%, and .3 pg/mg (pg/mg) defined the limit of quantification (LOQ).

2.5. Statistical analysis

2.5.1. Data preprocessing

HCC was log-transformed to correct positive skewness [55]. Following previous research on steroid hormones [56,57], extreme baseline HCC values beyond the 2.5th and 97.5th percentiles were excluded (n = 10). After excluding a missing HCC case (n = 1) and missing covariate data (n = 15), the final sample size for HCC analyses was n = 176 (OLP n = 85, control n = 91). A change score was calculated by subtracting baseline HCC (T0) from the intervention month HCC (T5).

For self-report measures (NA, PAF, DASS-S, VAS), the area under the curve with respect to increase (AUCi) was computed, a summarizing indicator that captures changes within the intervention month (T1 to T5). Each AUCi was residualized by regressing it on its baseline measure to control for potential baseline differences. Subsequently, a confirmatory factor analysis (CFA) was conducted to model the latent factor psychological distress defined by the four AUCi indicators. This approach allows us to estimate the shared variance across the four indicators, thereby improving construct validity and statistical power compared to separate analyses. The loading of NA AUCi was fixed to 1 for model identification. No missing data were detected, and one extreme outlier was excluded (>3 × interquartile range), resulting in N = 201 for analyses with psychological distress. The robust maximum likelihood estimation (MLR) was employed due to deviations from multivariate normality. Model fit was evaluated via Comparative Fit Index (CFI), Tucker-Lewis Index (TLI), Root Mean Square Error of Approximation (RMSEA), Standardized Root Mean Square Residual (SRMR), and the χ² statistic using conventional cutoff criteria (CFI/TLI ≥.95, RMSEA ≤.06, SRMR ≤.08; [58]). The latent factor scores were extracted for subsequent analyses. The CFA was performed using the lavaan package in RStudio (version 4.3.2).

2.5.2. Data analysis

The baseline values and demographic characteristics were tested for group differences using independent t-tests, Mann-Whitney-U-tests, or χ² tests, as appropriate. To examine the OLP effects (objective 1), two ANCOVAs were conducted with psychological distress and HCC as dependent variables and group (OLP vs. control) as the between-subject factor. Moderated regression analyses (objectives 2-4) were conducted to examine personality traits as potential moderators of OLP effects, with group as the independent variable (0 = OLP, 1 = control) and HCC or psychological distress as the dependent variable. Conscientiousness and neuroticism were primary moderators; extraversion, agreeableness, and openness were exploratory moderators. Continuous predictors were mean-centered. In case of a significant interaction, simple slope analyses are reported. Effect sizes for interaction terms were computed using Cohen's f² based on the change in explained variance between models with and without the interaction term. To compute the 95% confidence intervals (CI), bootstrapping with 5000 resamples and heteroskedasticity-consistent standard errors (HC3) were used.

Age, sex, and job hours (hours worked alongside academic studies) were included as a priori defined covariates to control for potential confounding based on associations with the outcome measures in previous studies [[59], [60], [61], [62]]. As psychological distress measures were associated with study semester, degree program, and significant stressors experienced in the past three months, these covariates were additionally included in the respective analyses. For HCC analyses, further a priori defined hair cortisol-related confounders (physical activity, body mass index, ultraviolet exposure, frequency of hair washes, hair sample weight, hormonal contraceptives, and season) and the baseline value were statistically controlled [55,62,63] (see Supplementary Material Tables S1–S6 for correlations with covariates). Analyses were conducted with IBM SPSS statistics 27.0, using two-tailed tests with α = .05.

3. Results

3.1. Sample characteristics and latent factor modeling

Demographic data and baseline values of participants are presented in Table 1, showing no significant differences between the groups before starting the trial (all p > .05, see Table S7 in Supplementary Material for the descriptive statistics in the total sample). Participants in the OLP group showed high adherence to the regimen, with an overall adherence of 96.8 ± 5.04%, which was not significantly correlated with the personality traits (all p > .05, see Table S8 in Supplementary Material). In addition to the assessed oral exam, all participants had prior (M = 1.7, SD = 1.2) and upcoming exams (M = 1.3, SD = 1.0). The number of additional exams did not differ significantly between the groups (p > .05).

Table 1.

Demographic and baseline data for OLP and control group.

Characteristics	OLP group	Control group	Test statistics	p	Effect size
n	100 (49.5%)	102 (50.5%)
Age (in years)	25.52 ± 4.10	26.53 ± 4.88	t(200) = −1.591	.113	d = .224
Females/males1	82/18 (82/18%)	75/27 (73.5/26.5%)	χ2(1) = 2.093	.148	Φ = .102
Study semester2	2.12 ± 1.42	2.14 ± 1.18	U = 4883	.582	r = .039
HCC (pg/mg)	4.17 ± 2.56	4.90 ± 3.50	t(189) = −1.222	.223	d = .177
Negative affect (NA)	1.50 ± .50	1.44 ± .56	t(200) = .736	.463	d = .104
Test anxiety (PAF)	45.30 ± 10.96	46.45 ± 11.63	t(200) = −.724	.470	d = .102
Subjective stress (DASS-S)	6.26 ± 4.03	7.06 ± 4.56	t(200) = −1.318	.189	d = .186
Exam-related stress (VAS)	59.98 ± 29.92	65.22 ± 28.66	t(200) = −1.270	.205	d = .179

Notes. Mean ± 1 standard deviation and frequency (percentage) are reported per group. Group differences at baseline were tested using t-tests, ¹χ² test, and ²Mann-Whitney-U-test. Test statistics, significance, and effect sizes are reported. HCC n = 176.

The CFA based on the four AUCi values (negative affect, test anxiety, subjective stress, and exam-related stress) indicated an excellent model fit to the data (χ²(2) = .048, p = .976; robust CFI = 1.000; robust TLI = 1.048; robust RMSEA = .000 (90% CI [.000, .000]); SRMR = .004). The standardized factor loadings were statistically significant (p < .001), ranging from .50 to .85, indicating moderate to strong associations between the observed AUCi indicators and the latent factor psychological distress. The factor demonstrated an acceptable internal consistency (composite reliability of .71, McDonald's ω = .71) and was thus carried forward for subsequent analyses.

3.2. OLP effects on HCC and psychological distress

The ANCOVA revealed significant group differences in psychological distress, approaching a medium effect size (F[1, 193] = 11.354, p < .001, ${\eta }_p^2$ = .056), indicating that the OLP group reported lower psychological distress during the intervention period (M_adjusted = −.178, SE = .074) compared to the control group (M_adjusted = .174, SE = .073). In HCC, a significant group effect was also found (F[1, 162] = 4.449, p = .036, ${\eta }_p^2$ = .027), indicating decreasing HCC levels in the OLP group (M_adjusted = −.053, SE = .024) and increasing HCC in the control group (M_adjusted = .017, SE = .023) during the intervention compared to the pre-intervention interval (see Fig. 2 and Supplementary Material Figs. S1–S2 and Table S9 for additional information on the intervention period). Partial correlation analyses revealed no significant association between changes in HCC and psychological distress in the total sample (r = .043, p = .593), nor within the OLP (r = −.101, p = .409) or control group (r = .091, p = .436).

Fig. 2.

Change in A psychological distress and B HCC during the OLP intervention month. Plots represent estimated marginal means of psychological distress (OLP n = 99, control n = 102) and HCC (OLP n = 85, control n = 91) derived from ANCOVAs. Error bars represents ±1 standard error of the mean. *p < .05, ***p < .001.

3.3. The moderating roles of conscientiousness and neuroticism

The interaction between group and conscientiousness in moderated regression analysis was significant (b = .331, SE = .161, t(191) = 2.055, p = .041, 95% CI = [.013, .649], f² = .018) and increased the explained variance in psychological distress (ΔR² = .016, F(1,191) = 4.222). Simple slopes analysis indicated that the effect of group on psychological distress was not significant at low levels (−1 SD) of conscientiousness (b = .159, SE = .147, t(191) = 1.083, p = .280). However, for individuals whose conscientiousness levels are average (i.e., at the mean), group significantly predicted psychological distress (b = .359, SE = .107, t(191) = 3.356, p = .001), and this effect was even stronger at high levels (+1 SD) of conscientiousness (b = .559, SE = .143, t(191) = 3.914, p < .001). These results show that the difference in psychological distress between the OLP and control group becomes more pronounced as conscientiousness increases, indicating stronger OLP effects on psychological distress at higher levels of conscientiousness. The moderation analyses with neuroticism revealed no significant interaction between group and neuroticism (b = −.005, SE = .180, t(191) = −.026, p = .979, 95% CI = [-.361, .351], f² < .001), and no improvements in the model (ΔR² < .001, F(1,191) = .001), indicating that the association between group and psychological distress is independent of neuroticism scores (see Fig. 3A).

Fig. 3.

Simple slope plots for moderated regression analyses showing the association between conscientiousness or neuroticism (mean centered) as moderators with A psychological distress and B HCC for OLP and control group. Shaded areas represent ±1 standard error around the regression lines.

Concerning HCC, the interaction between group and conscientiousness was not significant (b = −.071, SE = .081, t(160) = −.880, p = .380, 95% CI [–.230, .088], f² = .010; ΔR² = .007, F(1,160) = .774), suggesting that the relationship between group and HCC does not significantly vary depending on levels of conscientiousness. However, the interaction between group and neuroticism was significant (b = .132, SE = .063, t(160) = 2.080, p = .039, 95% CI [.007, .257], f² = .045), indicating a moderation effect that increased the explained variance in HCC (ΔR² = .031, F(1,160) = 4.326). The simple slopes analysis showed that the group effect on HCC was not significant at low levels (−1 SD) of neuroticism (b = −.018, SE = .047, t(160) = −.394, p = .694), whereas for individuals with average neuroticism, the simple slope was significant (b = .072, SE = .034, t(160) = 2.099, p = .037), and even stronger at high levels of neuroticism (+1 SD; b = .162, SE = .062, t(160) = 2.590, p = .010), indicating that the group effect on HCC increases as neuroticism rises (see Fig. 3B). Thus, the effect of OLPs on HCC becomes stronger at higher levels of neuroticism.

3.4. Exploratory analyses

Results of the exploratory moderated regression analyses revealed no significant interaction effects between group and extraversion, agreeableness, or openness in predicting psychological distress or HCC (p > .05, see Table 2). In sum, these exploratory findings suggest that OLP responses in psychological distress and HCC were independent of these personality traits.

Table 2.

Interaction effects between group and personality traits of moderated regression analyses in predicting psychological distress and hair cortisol.

Interaction	Dependent variable
	Psychological distress (n = 201)						Hair cortisol (n = 176)
	b	SE	t(191)	p	95% CI	f2	b	SE	t(160)	p	95% CI	f2
Group * Conscientiousness	.331	.161	2.055	.041	[.013, .649]	.018	−.071	.081	−.880	.380	[-.230, .088]	.010
Group * Neuroticism	−.005	.180	−.026	.979	[-.361, .351]	<.001	.132	.063	2.080	.039	[.007, .257]	.045
Group * Extraversion	−.202	.212	−.954	.341	[-.620, .216]	.006	−.040	.067	−.607	.545	[-.172, .091]	.003
Group * Agreeableness	−.018	.243	−.072	.942	[-.462, .497]	<.001	−.048	.089	−.540	.590	[-.224, .128]	.002
Group * Openness	.076	.170	.445	.657	[-.260, .412]	.001	−.014	.061	−.236	.814	[-.135, .106]	<.001

4. Discussion

The present study investigated the effects of a four-week OLP treatment on psychological distress and HCC in a real-life stress scenario—an oral university exam—and whether personality traits moderated these effects. Results indicated that OLPs reduced both psychological distress and HCC compared to a no-treatment control group. Moreover, individuals high in conscientiousness showed particularly strong reductions in psychological distress, while those high in neuroticism exhibited the largest decreases in HCC. No moderating effects emerged for extraversion, agreeableness, or openness.

Compared to the control condition, the OLP intervention significantly reduced psychological distress, approaching a medium effect size. This finding supports accumulating evidence that placebos without deception can alleviate psychological distress and extends prior OLP research by demonstrating effects over a four-week period [7,8]. Given the high prevalence of distress among university students and its associated health risks [64,65], this finding highlights the potential of OLPs for stress management in vulnerable populations. The magnitude of this OLP effect corresponds to meta-analytic evidence in self-report outcomes [10] and to effects reported for stress-management interventions in healthy populations [66].

Beyond subjective measures, participants receiving OLPs showed significantly lower HCC compared to controls, providing initial evidence that OLPs may modulate a long-term physiological stress marker under a powerful naturalistic stressor that elicits strong cortisol responses [22]. While prior OLP studies on salivary cortisol found no overall main effects [[14], [15], [16]], HCC reflects cumulative HPA axis activity and is less susceptible to diurnal or situational fluctuations [23]. It is linked to stress-related psychiatric symptoms, serious life events, and academic stress [62,67]. In the present sample, changes in HCC and psychological distress were not significantly correlated, which is consistent with the generally weak or inconsistent associations between subjective and physiological stress indicators (e.g., Ref. [20]) and suggests that they capture partly distinct dimensions of stress responses. Hair integrates cortisol responses over several weeks, whereas the self-reports capture momentary and short-term experiences [68]. Consequently, HCC reflects a complementary, temporally distinct dimension of stress reactivity. The OLP effect on HCC and psychological distress extends our previous work, which focused on a different research question, and supports the validity of the initial findings on psychophysiological OLP effects [69] in an expanded sample from the same longitudinal RCT.

Analyses on the role of personality traits revealed a dissociation. Conscientiousness moderated the OLP effect on psychological distress but not on HCC, whereas neuroticism moderated the OLP effect on HCC but not on psychological distress. This suggests that personality traits may differentially shape subjective versus physiological stress responses, as observed in other studies [27,70,71], resulting in outcome-specific variations of the traits’ influence on OLP responsiveness.

The group differences (OLP vs. control) in psychological distress increased with higher conscientiousness, expectedly indicating conscientiousness as moderator of responsiveness to OLP interventions in academic contexts. Conscientiousness has not previously been examined as moderator of OLP effects yet, and prior studies on deceptive placebos found no substantial associations between conscientiousness and placebo responsiveness in pain paradigms [18,24]. However, the present study examined performance-related, social-evaluative distress – an environment particularly salient for conscientious individuals, emphasizing that personality effects on placebo responsiveness are context-dependent (e.g., stress vs. pain) [27,28]. In such contexts, conscientious individuals may experience heightened distress, potentially due to their high investment in goal attainment and strong attachment to previously defined goals. Consistent with this, the control group in the present study showed increased psychological distress in highly conscientious individuals, aligning with previous evidence [35,36].

Interestingly, these highly conscientious individuals benefited more strongly from the OLP intervention, resulting in decreased psychological distress. Consistent with this, an RCT showed increased academic stress in highly conscientious students, whereas conscientious students who received a mindfulness training showed higher stress reductions [72]. Conscientious individuals tend to achieve more favorable outcomes in psychotherapeutic and pharmacological treatments [25], aligning with the stronger OLP effect in psychological distress. Given the association between conscientiousness and treatment adherence [37], conscientious individuals may have been more likely to engage with the OLP instruction. Overall, conscientiousness may entail both vulnerability to psychological distress in performance contexts and enhanced responsiveness to OLP intervention addressing these stressful experiences.

Conscientiousness did not significantly moderate OLP effects on HCC. This might be consistent with meta-analytic findings that conscientiousness is more consistently linked to subjective than to physiological stress responses [73]. However, this does not necessarily imply that conscientiousness is unrelated to physiological stress processes. Erickson et al. [42] demonstrated that conscientiousness did not exert a direct effect on HCC, but moderated the association between neuroticism and HCC, buffering neuroticism-related increases in cortisol levels. Given that conscientiousness is associated with self-regulatory behaviors [34] and adaptive coping strategies [74], it may exert its influence on physiological stress processes in an indirect, interaction-based manner, particularly by attenuating the impact of affective traits on HPA-axis activity rather than manifesting as a stable main effect. These self-regulatory strategies thus may facilitate adaptive regulation of physiological stress responses [75], without necessarily diminishing the subjective stress perception [76], potentially explaining why conscientiousness’ influence on OLP responses emerged only in psychological distress.

Neuroticism significantly moderated the OLP effect on HCC, with group differences becoming more pronounced as neuroticism scores increase. In the control group, highly neurotic individuals showed elevated HCC during the exam preparation period, consistent with evidence linking neuroticism to heightened distress during academic stress [45] and stronger physiological stress responses, including elevated salivary and hair cortisol [42,43] and higher cardiovascular reactivity [41].

Within the OLP group, higher neuroticism predicted stronger reductions in HCC, indicating enhanced physiological responsiveness to the OLP treatment. This aligns with research on deceptive placebo responses, which similarly implicate neuroticism in modulating placebo effects [27,46,77]. Highly neurotic participants responded more strongly to a placebo believed to contain caffeine, reflected in improved cycling performance [46]. Likewise, placebo analgesia has been shown to be positively associated with neuroticism [77], although findings in pain domains are mixed [17,24]. Of particular relevance to our study, Darragh et al. [27] demonstrated that higher neuroticism was related to enhanced physiological placebo responses in a psychosocial stress context, evidenced by reduced heart rate and increased heart rate variability, indices that mirror autonomic nervous system (ANS) activity. The ANS and HPA axis interact in stress responses [78] and are both linked to neuroticism [41,42]. Building on this, our results suggest that neuroticism is a relevant moderator of physiological stress and may shape placebo responses in a long-term marker of HPA-axis activity, even in non-deceptive paradigms.

The moderation effect of neuroticism did not emerge in OLP effects on psychological distress. Given that neuroticism is associated with self-reported distress [39,45,73], we initially expected higher neuroticism would also moderate OLP response in psychological distress, an effect that was not observed in our results. Nevertheless, as depicted in Fig. 3A, higher neuroticism was descriptively related to increased psychological distress, but the interaction between group and neuroticism was not significant, indicating that the beneficial effect of OLPs on psychological distress was independent of the traits’ level.

Neuroticism has been linked to maladaptive coping, characterized by less cognitive reappraisal and greater emotion-focused strategies [74], both linked to stronger physiological reactivity [75]. In this context, the OLP intervention may have supported more adaptive coping [8], thereby improving physiological stress regulation in highly neurotic individuals, reflected in reduced HCC. At the same time, self-reported distress remained elevated and reduced in variability due to neuroticism's relatively stable disposition to negative affectivity and conceptually close relation to psychological distress [79]. Thus, physiological long-term indicators such as HCC may more sensitively capture changes in stress regulation, while neuroticism-related placebo responses in stress contexts emerged primarily at the physiological level. Similar dissociations have been observed previously, showing no significant association between neuroticism and placebo effects in self-reported stress, while placebo responses in physiological stress measures were associated with neuroticism [27]. Together, OLPs may counteract the physiological hyperreactivity associated with neuroticism even when psychological distress remains unaffected.

Exploratory analyses showed no significant moderating effects of extraversion, agreeableness, or openness on OLP responses in either psychological distress or HCC. This aligns with investigations on personality predictors of OLP responsiveness [29,30] and with deceptive placebo studies reporting non-significant effects of these traits (e.g., Ref. [7]). Although these traits are generally associated with treatment responses in therapeutic contexts [25], they may influence responsiveness only under specific interpersonal conditions. Kelley et al. [24] demonstrated that extraversion, agreeableness, and openness were associated with placebo responses primarily in warm, empathic clinical interactions, whereas such effects disappeared in neutral or minimal-contact contexts. Because experimenter–participant interactions were deliberately standardized and neutral in the present study, interpersonal cues that could activate or amplify the relevance of such traits were minimized, possibly explaining the null effects. Future studies should vary the interpersonal context of OLP administration to test whether these traits become more influential under socially engaging conditions.

Several limitations should be acknowledged. Attrition due to exam cancellation may have introduced selection bias, potentially underrepresenting highly distressed students. Furthermore, self-selection may have resulted in participants who were more open to placebo-based interventions, more motivated by distress to seek any intervention, or held more favorable expectations, potentially limiting external validity. However, prior OLP studies indicate that expectancy levels are generally low [80] and their associations with OLP effects are inconsistent [[81], [82], [83]], suggesting that such factors may be less likely to explain the observed effects.

The naturalistic academic stressor enables ecologically valid observations of OLP effects in university students and facilitates comparisons with similar OLP studies [7,8]. The oral exams only varied across health-related study programs, thereby enhancing the generalizability of the findings across different academic disciplines, while also sharing similar assessment formats (e.g., similar timeframe, one examiner). The number of prior and upcoming exams was low, limited in variance, and did not differ significantly between the groups, indicating a comparable number of additional exams across conditions. Furthermore, the number of additional exams was not significantly associated with HCC or psychological distress within groups. In sum, these findings suggest that these contextual factors are unlikely to have contributed to the observed effects. Future research should extend this work to other populations and stressors.

Although biological sex was controlled for, the unequal distribution in the present sample (77.6% female) may limit generalizability. Moreover, including a deceptive placebo condition in future studies could further contextualize OLP effects relative to non-specific factors. Although participant contact was standardized across groups, reducing the likelihood of differences in perceived support or attention, it remains less clear to what extent effects reflect OLP-specific mechanisms versus non-specific factors such as self-monitoring or demand characteristics (e.g., Hawthorne effect). This limitation arises from the inability to blind OLP interventions.

Future research may benefit from assessing more fine-grained behavioral and cognitive aspects related to the intervention (e.g., daily diligence in following the intervention regimen and engagement with the rationale) to further clarify the relationship between personality traits and OLP effects. Finally, this study was the first to examine both physiological and psychological OLP responses in relation to personality traits, warranting replication in adequately powered studies to further validate the observed effects. Given the individual testing approach based on assumptions about personality effects [84], future studies may benefit from pre-registering hypotheses to further enhance transparency. This research project will extend these investigations by addressing different research questions that examine the OLP effect on additional outcome measures (e.g., exam performance, acute affective and physiological stress responses) and psychological factors that may moderate OLP effects in naturalistic stress contexts.

In sum, the present study underscores that the efficacy of OLP interventions may be trait-dependent, suggesting that certain individuals - particularly those more vulnerable to academic distress - may derive the greatest benefit from OLP treatments. Beyond demonstrating that OLPs can impact an objective biomarker, the study extends the scope of ethically acceptable placebo interventions as a potentially valuable tool for mitigating negative stress-related consequences in non-clinical populations.

Data availability

The referenced dataset that supports the findings of this study are available on the Open Science Framework (OSF), https://osf.io/9whjx.

Funding sources

This project was funded by the Deutsche Forschungsgemeinschaft (DFG, German Research Foundation) – Project No. 505631897, EN 1155/3-1.

CRediT authorship contribution statement

Carolin Liedtke: Data curation, Formal analysis, Investigation, Methodology, Visualization, Writing – original draft, Writing – review & editing. Michael Schaefer: Conceptualization, Funding acquisition, Methodology, Project administration, Supervision, Writing – review & editing. Sören Enge: Conceptualization, Funding acquisition, Methodology, Project administration, Supervision, Writing – review & editing.

Declaration of competing interest

The authors declare the following financial interests/personal relationships which may be considered as potential competing interests:Soeren Enge reports financial support was provided by German Research Foundation. If there are other authors, they declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.

Appendix A. Supplementary data

The following are the Supplementary data to this article: