Introduction

Meditation is the process of training attention and awareness to achieve a physiological state that elicits physical and mental relaxation and enhances emotional stability (Bajaj et al., 2019; Jevning et al., 1992; Lee et al., 2015; Young & Taylor, 1998). Together with the associated state of mindfulness—being non-judgmentally focused wholly on the present moment—meditation practice has been found to have positive effects on mental and physical health (Arias et al., 2006; Grossman et al., 2004; Hofmann et al., 2010; Lutkajtis, 2019a). With respect to mental health, positive effects have been observed mainly for reducing symptoms of stress and depression, although benefits for anxiety have been observed when meditation or mindfulness are practiced in conjunction with other treatment approaches (Saeed et al., 2019). Observed physical benefits include improved sleep quality (Rusch et al., 2019), and reduced blood pressure and heart rate, implying reduced risk of cardiovascular disease (Conversano et al., 2021; Kayloni & Emery, 2015; Krittanawong et al., 2020; Levine et. al, 2017).

Reviewers have consistently called for increased theoretical and methodological rigor when exploring mechanisms underlying the effects of meditation (Hickey, 2010; Lomas et al., 2015; Ospina et al., 2007; Younge et al., 2015), especially with respect to cardiovascular functioning, where meta-analyses of the direction of effect have shown mixed results (Conversano et al., 2021; Dawson et al., 2020; Olex et al., 2013; Schnaubelt et al., 2019; Scott-Sheldon et al., 2020; Zhang et al., 2021). Proposals regarding mechanisms linking meditation and cardiovascular functioning focus on stress reduction. Stress is associated with continued activation of the sympathetic (‘fight-or-flight’) nervous system without counteraction by the parasympathetic (‘rest-and-digest’) nervous system. Long-term stress results in hormonal imbalances and maladaptive lifestyle patterns (e.g., alcohol use, lack of exercise, and poor sleep) that represent risk factors for hypertension and cardiovascular disease (Conversano et al., 2021; Won & Kim, 2016). Arguably, meditation can reduce sympathetic nervous system hyperactivation because it increases parasympathetic activity while reducing sympathetic activity or leaving it unchanged (Gerritsen & Band, 2018; Kozhevnikov, 2019; McCorry, 2007; Shr-Da & Pei-Chen, 2008). However, in what is known as the ‘meditation paradox’, it has been observed that, in some individuals, certain relaxation and meditation techniques produce increased heart rate—an indication of sympathetic activation (Peng et al., 2004). As highlighted in a narrative review, neuroimaging patterns and subjective reports also suggest increases in general arousal and alertness following some forms of meditation (Britton et al., 2014).

In suggesting that some people might not experience parasympathetic dominance during meditation, the meditation paradox is consistent with the observation by some reviewers that theoretical perspectives on the health benefits of meditation lack clarity with respect to possible adverse effects of meditation practice (Lauricella, 2014). A recent narrative critique of literature on the health benefits of meditation emphasizes that only a small number of studies have explored adverse effects, largely due to a broader motivation in the media and the scientific community to present meditation as a ‘simple’ side-effect-free solution to various health problems (Lutkajtis, 2019a). Similarly, systematic reviews indicate that few trials of mindfulness-based therapies track adverse effects (Wong et al., 2018), partly because there are currently no requirements to report them or guidelines for doing so (Farias et al., 2020).

A documented negative side-effect of one type of meditation, Ānāpāna smrti, is nausea: a complex reaction of different physiological and psychological systems manifesting as epigastric discomfort combined with an urge to vomit and associated with increased sympathetic activity (LaCount et al., 2011; Muth et al., 1996; Sclocco et al., 2016). This side-effect has been discussed in ancient and contemporary meditation manuals, as well as qualitative reports from experienced meditators (Lindahl et al., 2017). In our own study of Ānāpāna smrti meditation, 17% of participants reported dizziness, nausea, extreme sweating, hot flushes, weakness, and faintness during meditation (Kotherová, 2015).

In the present study, we investigate whether feelings of nausea during Ānāpāna smrti meditation reflect increased sympathetic activity. The reverse pattern—an association between nausea and increased parasympathetic activity—is also possible, given that nausea often accompanies vasovagal syncope, a benign temporary decrease in blood flow to the brain caused by parasympathetic activation and observed during meditation (Alboni, 2015; Goshvarpour & Goshvarpour, 2021).

Our protocol involves measuring heart-rate-variability (HRV) indices of sympathetic and parasympathetic activation, as well as blood flow indicators, during three phases: (1) before, (2) during, and (3) after Ānāpāna smrti meditation. Feelings of nausea were assessed at the end of the third phase (i.e., post-meditation), and the analysis focuses on analyzing associations between the degree of reported nausea and the degree of change in heart-rate and blood flow parameters as participants moved from a resting state (pre-meditation) to the meditation phase and to rest again (post-meditation).

One notable strength of our protocol is that we recruit participants without prior meditation experience. Past studies of autonomic nervous system activity during meditation have tended to focus on experienced practitioners (Britton et al., 2014; Gerritsen & Band, 2018; Peng et al., 2004).

A further strength is that our choice of meditation technique is theoretically motivated. This is critical, given that different philosophical traditions, ritual practices and techniques for meditation have been shown to produce different physiological and behavioural effects (Brook et al., 2013; Kozhevnikov, 2019; Krygier et al., 2013; Lehrer et al., 1999; Levine et al., 2017; Lutz et al., 2008; Raffone et al., 2019; Rubia, 2009; Tang et al., 2009). We explore the effects of a meditation technique known to heighten parasympathetic activity (Kozhevnikov, 2019; Ñāṇamoli, 1998). Ānāpāna smrti (san.), also known as Ānāpāna sati (pál.), belongs to a group of Theravada Buddhist meditation techniques commonly known as Samadhi or Shamatha (Ñāṇamoli, 1998). Samadhi practice involves training concentration, and so practitioners are instructed to direct their “undivided attention” to the single object of meditation while withdrawing their focus from other objects (Buddhaghosa, 2010). Ānāpāna smrti practice is usually translated as “breath (ānāpāna) awareness (smrti)”. Instructions specifically concern mindful observation of breath inhalation and exhalation. During the meditation, the person focuses their mind on a singular object or process with eyes closed and trying to be aware as much as possible of internal bodily processes.

Overall, in investigating autonomic nervous system activity potentially responsible for nausea during Samadhi meditation, we hypothesize that higher levels of nausea will be associated with higher levels of activity in the sympathetic arm of the autonomic nervous system during meditation as compared to rest. Autonomic nervous system activity will be measured using HRV indices, but blood pressure (blood flow) will also be monitored during meditation and rest to explore known associations between nausea and increased parasympathetic activity, as reflected in increased probability of vasovagal syncope.

Method

Participants

Fifty-seven undergraduate students enrolled in an undergraduate university-wide research methods course (42 women; mean age = 22.6; SD = 1.7) participated in the study in exchange for course credit. All students enrolled in the course (approximately 150) had to complete a preliminary screener survey, and the participants in our study were invited to participate based on their screener survey responses. To be eligible for the study, students had to report no prior experience with any kind of meditation or yoga practice. Participants also had to report absence of a current mental health disorder, chronic heart condition or history of psychoactive drug use. For more detailed information about the participants, see Table 1 in the Results.

Table 1 Preliminary descriptive analysis comparing nausea groups
Full size table

Procedure

All participants provided written informed consent to participate in a study investigating physiological changes associated with cognitive load in a mental exercise. The word “meditation” was not mentioned at any point during recruitment and data collection to avoid creating expectations around what the effect of mental exercise should be (Lustyk et al., 2009).

The experiment took place at the Department of Internal Medicine and Cardiology at XY University Hospital. Participants were asked (a) not to consume alcoholic beverages, drugs or energy drinks for at least 24 h before taking part in the study, (b) not to consume caffeinated beverages for at least six hours before, and (c) refrain from eating for at least 2 h and follow a sufficient drinking regime up to 3 h before. During the entire procedure, participants were under supervision of a doctor and nurse. Testing took place in a quiet private room.

After being fitted with monitoring devices (see Measures: Equipment below), each participant underwent three phases in a set order: the pre-meditation phase (Phase 1; T1), the meditation phase (Phase 2; T2) and the post-meditation phase (Phase 3; T3). During the pre-meditation phase, the participant was instructed to sit quietly with their eyes open for five minutes while all heart-rate measures were collected. During the meditation phase, the same measures were collected during a 10-min meditation exercise described in the ‘Meditation’ section below. The post-meditation phase was identical to the pre-meditation phase.

Meditation

At the beginning of the meditation phase, participants were provided written instructions to sit quietly with their eyes closed and focus on their breathing by counting when breathing in and breathing out. Participants were further instructed to begin counting from the beginning if they found their attention being drawn away. The instructions were as follows:

When prompted, please close your eyes and keep them closed until told otherwise. Focus your attention on your breath. Focus on your inhalation and exhalation. While focusing on your breath, begin to count in your mind as follows: on inhalation say 1, on exhalation say 1. On the next inhalation say 2, and on the next exhalation say 2. Continue like this: 3–3, 4–4, and so on. Try not to lose your focus on the breath. If you lose focus (e.g., if you start following another thought or focus on some discomfort or emotion), start counting from the beginning; so, again, from the number 1 (1–1, 2–2 etc.). If you are unsure about any aspect of these instructions, feel free to ask the research assistant.

Participants were further informed that the task was not a race to the highest breath count, but an exercise in being present with the breath as much as possible. That is, they were informed that the goal of the task was not to reach the highest count, but to focus on breathing, seeing the counting as an aid to that process. Participants performed the meditation task for 10 min.

Measures

Equipment

A continual ECG Holter monitoring device (MARS 5000, GE Marquette Medical Systems, Milwaukee, United States) was used for measuring heart rate. A Portapress Model 2 (Finapres Medical System B.V. Amsterdam, The Netherlands), was used for non-invasive continuous blood pressure measurement. Additionally, outside the scope of the present study, a patient monitor (Dash 4000, GE Healthcare, Illinois, United States) was used for monitoring of breathing rate and blood oxygen saturation.

Heart Rate Variability Indices of Autonomic Nervous System Activity

Among the HRV measures, the Deceleration capacity (DC) of the heart rate served as an indicator of parasympathetic activity (Bauer et al., 2006). Normalized power in the low-frequency range (LF; 0.04–0.15 Hz; Malik, 1996) served as a marker of sympathetic activity, while normalized power in the high-frequency range (HF; 0.15–0.4 Hz; Malik, 1996) served as an additional marker of parasympathetic activity. The ratio of LF to HF (LF/HF) was calculated as a marker of sympathetic system dominance (i.e., of whether any sympathetic activation is larger in magnitude than parasympathetic activation; Kiyono et al., 2017).

Blood-Pressure Indices of Parasympathetic Nervous System Activity (Vasovagal Syncope)

Instances of vasovagal syncope in Phases 2 (meditation) and 3 (post-meditation) were identified by two independent cardiologists (the fourth and last authors, blind to participant allocation to nausea groups), who discussed conflicting judgements and came to a consensus. To arrive at the judgments, the cardiologists viewed, for each participant in each phase, graphs of mean arterial pressure plotted concurrently with beats-per-minute (see Part 2 of the Online Supplement for the graphs and judgements; DeMers & Wachs, 2019). For each participant, presence of vasovagal syncope in Phase 2 (meditation) or Phase 3 (post-meditation) was, finally, operationalized as a binary variable: 0 (No) or 1 (Yes).

Nausea Symptoms

Nausea symptoms were measured (at the end of Phase 3: post-meditation) using a Nausea Profile (Muth et al., 1996), a measure in which participants were instructed to indicate the extent to which they were feeling each of the following 17 symptoms on a scale from of 0 (not at all) to 10 (severely): “shaky”, “upset”, “lightheaded”, “sick”, “sweaty”, “queasy”, “worried”, “hopeless”, “fatigued/tired”, “panicked”, “nervous”, “scared/afraid”, “ill”, “discomfort in my stomach”, “might vomit”, “weak”, and “hot/warm”. In a back-translation process, the Nausea Profile was translated from English to Czech and then back to English by two independent translators and the final translation was compared to the original by a third person. The final nausea score was the sum of all responses, expressed as a percentage of the maximum possible score (170), in line with Muth et al. (1996)’s specifications. Thus, scores could range from 0 to 100, and reflected a well-established multifaceted conceptualisation of nausea as a composite of three correlated latent factors: somatic distress, gastrointestinal distress, and emotional distress (Bedree et al., 2023; Sanger & Andrews, 2023; Tarbell et al., 2023).

Additional Measures: Experiences During Meditation

At the end of Phase 3 (the post-meditation phase), participants were asked about different aspects of their meditation experience, including the extent to which they found the task demanding (“Please indicate how difficult you find the breath counting exercise:___” -3 = not at all, 3 = extremely). Participants were also asked about the extent to which they felt uncomfortable during the task (“To what extent during the breath counting exercise did you feel uncomfortable?” -3 = not at all, 3 = extremely), and the extent to which they felt they were successful at the task (“Please indicate how successful you think your breath-counting practice was:__”, -3 = not at all, 3 = extremely). Participants were also asked to estimate the number of counting errors they made during the task (“Try to estimate how many times you had to go back to count: __”), and to report the maximum number of breaths they were able to count during the task (“What is the highest number you were able to count to without interruption? __”).

To enable tracking of study artifacts, participants completed a final open-ended question, in which they were asked to describe what they experienced during the meditation (“Describe as accurately as possible how you experienced the experiment”). No participants were excluded on the basis of this description.

Statistical Analysis

To address the main research question regarding the relationship between HRV indices of autonomic (sympathetic and parasympathetic) activity and nausea, we fitted eight mixed effects models – two for each HRV outcome of interest: DC, LF, HF and the LF/HF ratio. (Fig. 1) The models were informed by the fact that we observed two grou**s of nausea scores—low and high (see Results: Descriptive statistics). In the models, the fixed effects were phase/timepoint (two levels: Phase 2 meditation vs. Phase 1 pre-meditation in one model, and Phase 2 meditation vs. Phase 3 post-meditation in the other model), nausea symptoms, nausea group (two levels: high vs. low), and all possible two-way interactions. The random effect was participant ID. Effectively, the two models for each outcome differed only in terms of the phases/timepoints they compared. Models were fitted using the lme4 package in R v. 4.2.3 (Bates et al., 2014). Descriptive graphs capturing phases/timepoints as difference scores (T2 minus T1 and T2 minus T3; i.e., Phase 2 minus Phase 1 and Phase 2 minus Phase 3) were drawn to assist with interpretation (see Fig. 2). Sample analysis code is available in Part 1 of the Online Supplement. Two model coefficients were of interest, given the research question: the timepoint x nausea symptom score interaction (indicating whether the degree of HRV change from rest to meditation would change depending on nausea score), and the timepoint by group interaction (indicating whether the degree of HRV change from rest to meditation would change depending on whether a participant’s nausea symptom score was, at a coarse-grained level, low or high).

To examine associations between nausea and increased parasympathetic activity, as reflected in increased probability of vasovagal syncope, we fitted a logistic regression in which presence (yes/no) of vasovagal syncope in Phases 2 (meditation) or 3 (post-meditation) was the outcome variable, and the predictors were nausea group, nausea symptom score, and an interaction term. Thus, we allowed for the possibility that the strength of the association between nausea symptoms and probability of syncope would differ by nausea group.

The characteristics of the two nausea groups on HRV indices at baseline and additional available measures of meditation experience were explored in a preliminary descriptive analysis involving independent-samples t-tests and Chi-square tests (see Results: Descriptive statistics).

Results

Descriptive Statistics

As shown in Fig. 1—a stem-and-leaf plot of nausea scores—we found that nausea symptom scores formed two clusters, in that no scores fell between 37 and 42. Based on this finding and the descriptive scatter plots in Fig. 2, we modelled the HRV-nausea relationship separately for the two clusters. In a finding that provides support for this decision, Muth et al. (1996) found 42 to be the average nausea level experienced during optokinetic rotation by people prone to motion sickness. Table 1 summarises how the clusters compared with respect to demographics, meditation experience, and baseline HRV indices. High scorers (n = 12) differed from low scorers (n = 45) in that they were all women, more likely to find meditation demanding and uncomfortable, more likely to make errors during meditation, and likely to count to a lower number during meditation.

Fig. 1
figure 1

Stem-and-leaf plot of nausea scores, with the cut-offs for the identified nausea groups shaded in red

Full size image
Fig. 2
figure 2

Descriptive statistics for HRV difference scores–scatter plots, means (larger dots), and slopes indicating correlation strength

Full size image

The HRV-Nausea Relationship

The results of the mixed-effects modelling assessing the relationship between nausea symptoms and meditation-related changes in each of the HRV indices – DC and HF (markers of parasympathetic activity) and LF and the LF/HF ratio (markers of sympathetic activity) – are presented in Table 2. Figure 2 plots the associated descriptive statistics. Results indicate that, in line with our hypothesis, higher nausea, in terms of group membership, was associated with increased sympathetic activity, in that the timepoint x group interaction effect was significant and positive for both models for LF and the second model (of the last two timepoints: T2 and T3) for the LF/HF ratio. In line with the previously observed relationship between higher nausea and vasovagal syncope, a marker of parasympathetic activity, we also observed a positive correlation between HF—a marker of parasympathetic activity – and nausea in both associated models (for T2 and T1, as well as T2 and T3). Notably, across both groups, and models for both the first two timepoints and last two, we observed non-zero change levels in DC (i.e., a main effect of timepoint), indicating an increase in parasympathetic activity during meditation.

Table 2 Results and interpretation of the mixed effects modelling
Full size table

The Syncope-Nausea Relationship

To directly assess the relationship between higher nausea and the probability of vasovagal syncope, we conducted a logistic regression in which presence of vasovagal syncope in Phase 2 (meditation) or 3 (post-meditation) was the outcome variable, and the predictors consisted of nausea symptom score, nausea group, and the group x score interaction. In the low nausea group, 20 participants (48%) were identified as having experienced vasovagal syncope at least once across Phases 2 and 3. Thus, in terms of descriptive statistics, results did not support the hypothesis, since syncope was less (rather than more) likely in the high nausea group. In the logistic regression, as Table 3 shows, neither group membership, nor nausea score, nor the combination of group membership and nausea score predicted the probability of syncope.

Table 3 Coefficients in the logistic regression of vasovagal syncope in T2 or T3 on nausea symptom score, nausea group, and the interaction term
Full size table

Discussion

Calls for theoretically guided investigations of cardiovascular functioning during various specific types of meditation have been made in light of observations of adverse effects of meditation in some individuals (e.g., Lauricella, 2014). The present study investigated heart rate variability and blood pressure (specifically, vasovagal syncope) in people practicing Samadhi meditation for the first time and reporting varying levels of the adverse side-effect of nausea. Overall, compared to a resting state, meditation was found to increase parasympathetic activity, although reports of high—compared to low—levels of nausea were associated with greater sympathetic activity during meditation. Meanwhile, in both the ‘low’ and ‘high’ nausea clusters, incremental increases in nausea were associated with incremental increases in parasympathetic activity. Further investigation involving blood pressure indicated that this association at the incremental level was not due to the effects of vasovagal syncope.

It might follow from our findings of correlated incremental increases in parasympathetic activity and nausea that, for people with an overactive sympathetic nervous system, Samadhi meditation—known for its activation of the parasympathetic nervous system—might produce a stronger parasympathetic response. To use a crude analogy, people who drive faster all the time (i.e., have an overactive sympathetic nervous system) also break harder all the time (i.e., are prone to stronger parasympathetic responses). We did not have the statistical power to investigate the effects of the interaction between nausea score and nausea group (i.e., cluster) on change in parasympathetic activity over time. A study powered to investigate such an interaction (through adding a three-way interaction to our mixed modelling framework) would be a logical next step.

Extending on dominant reviews in the area, our findings showed that activation of the sympathetic nervous system can be present not only in Vajrayana types of meditation (Kozhevnikov, 2019) but also in Samadhi meditation, which is the basic technique underlying mindfulness practice (Nilsson & Kazemi, 2016). Additionally, our study shows that, despite the parasympathetic system's general activation during Samadhi meditation, sympathetic nervous system activation and associated adverse physical symptoms can occur in some individuals (Farmer et al., 2015; Lutkajtis, 2019a;). This finding echoes Britton et al.’s (2014) observation in a narrative multidisciplinary review that meditation is a complex phenomenon that can elicit both parasympathetic and sympathetic nervous system activation—hypoarousal and hyperarousal—depending on dose, expertise, and individual factors.

On a practical level, our findings raise questions about whether meditation is ‘for everyone’. Our study identified adverse effects (nausea resembling motion sickness in terms of severity) in one out of five novice practitioners—12 out of 57 (21%). This is higher than the 11% rate of incidence of gastrointestinal symptoms reported in a recent systematic review of experimental, observational, and case-study-based evaluations of mindfulness practice (Farias et al., 2020).

People with high anxiety might stand to benefit most from the autonomic rebalancing meditation potentially offers, but, given the known proneness of more anxious people to nausea (e.g., Haug et al., 2002), longitudinal investigations are needed to determine whether people with high anxiety and stress levels tend to overcome the nausea barrier identified in the present study. Participants in the ‘high nausea’ group reported more challenges with meditation (i.e., greater demands and discomfort, and lower perceived success), so there is evidence that people are conscious of initial barriers, and could benefit from explicit support in overcoming them.

More broadly, our findings highlight that research into meditation and its utilization in healthcare should be based on a comprehensive understanding of the effects of various meditation techniques and their particular constituent elements. Despite pressure to characterize meditation as a low-cost intervention with widespread benefits, researchers must not lose sight of individual differences in physiological and psychological responding (Lutkajtis, 2019a, 2019b). Moreover, adverse effects should be documented when conducting trials of easy-to-access meditation and mindfulness apps (Taylor et al., 2022).

Study Limitations

A key limitation of our findings is that we did not have a priori expectations of two nausea clusters, and thus had limited measures on which to compare and distinguish the ‘low’ and ‘high’ nausea groups. Clearer characterization of these groups is a task for future research. In light of recent findings identifying decreasing parasympathetic activity with increasing nausea in women but not men (Caillet et al., 2022), future research should seek to overcome the confounding effects of gender—effects that might have been present in the current study, given the gender imbalance across groups. Overall, we present findings that should inform future study designs and power analyses. We did not conduct a power analysis of our own, but our sample size exceeds or equals those of previous studies on adverse effects of meditation (Lindahl et al., 2017; Shapiro, 1992), and it was reassuring to see observed effects replicating across models in the ‘first two’ and ‘last two’ phases—models that differed only in terms of whether the baseline occurred before or after meditation.

A further limitation of our work is that we observed a positive relationship between nausea levels and HF, but not the other measured marker of parasympathetic activity—DC. Our findings are consistent with emerging evidence that the relationship between HF and DC is imperfect (Lewek et al., 2009; Pan et al., 2016), particularly at lower breathing rates, such as those likely induced by the meditation task in the present study (Wang et al., 2016). Future research should measure and report participants’ breathing rates.

Another potential limitation of our design is that, in exploring the relationship between meditation and parasympathetic activity, we focused on vasovagal syncope rather than smaller and more volatile changes in blood pressure and heart rate. Techniques for detecting these more fine-grained changes are emerging (van Dijk et al., 2020), and might allow for increased precision in capturing parasympathetic activity in future meditation research.

Conclusions

In conclusion, we show that Samadhi meditation, which forms the basis for mindfulness practice, produces marked nausea symptoms in one-fifth of people attempting it for the first time. In line with findings that nausea can reflect both sympathetic (fight-flight) and parasympathetic (rest-digest) activation, we observed heightened heart-rate variability indices of sympathetic activation among the one-fifth of participants reporting stronger nausea, but also, in both groups, increasing HRV indices of parasympathetic activity with increasing reported nausea strength. We did not, however, observe increasing probabilities of vasovagal syncope (a blood-pressure-based index of parasympathetic activity) with increasing nausea levels. A future study would need to investigate whether some people (with heightened nausea and sympathetic activation) are prone to stronger parasympathetic (braking) responses. However, more broadly, our findings demonstrate how critical it is to monitor the potential negative side-effects of meditation, as well as the mechanisms contributing to them.