Something Old, Something New, Something Borrowed, Something Green: How Different Domains of Traditional, Alternative, and Complementary Medicine Use Are Rooted in an Irrational Mindset

There is no definite agreement on what should be considered TCAM (Wieland et al., 2011), despite there being over 100 heterogeneous healing philosophies, therapeutic modalities, and diagnostic methods treated as TCAM practices (Ernst, 2019). Conceptual taxonomies group TCAM practices on the basis of similarities in their approaches and beliefs on health, treatment, and healing (e.g., Barnes et al., 2004; Kaptchuk & Eisenberg, 2001; Wieland et al., 2011), usually differentiating between several overlapping domains: (1) alternative medical systems representing complete systems of theory and practice that evolved parallel to conventional medicine (e.g., acupuncture, Ayurveda, homeopathy); (2) mind-body medicine emphasizing the impact of thoughts and emotions on physical health (e.g., meditation, guided imagery, hypnosis); (3) natural product-based therapies stemming from the idea that substances and functional foods found in nature have health benefits over and above basic nutrition; (4) manipulative and body-based practices involving manipulation or movement of body parts (e.g., chiropractic, osteopathy, reflexology); (5) energy medicine, typically incorporated into New Age healing, which uses energy fields for healing (e.g., Reiki, therapeutic touch, crystal therapy, spiritual healing); and (6) traditional medicine, including diverse parochial folk medicine practices indigenous to a given culture (e.g., copper bracelet for arthritis, cabbage leaves for inflammation), as well as religious healing.

Although these theoretical groupings provide a useful conceptual framework, they may or may not correspond to actual patterns of TCAM use. To the best of our knowledge, there have been only two attempts to derive an empirical taxonomy of TCAM use. Ayers and Kronenfeld (2010) found that 20 TCAM practices group around four latent dimensions: alternative medical systems (including folk and herbal medicine), mind-body treatments, prayer, and manipulative treatments. Drawing from a sample of breast cancer patients, Ashikaga et al. (2002) found that 11 TCAM practices grouped around two factors. Since previously detected empirical structures of TCAM relied on a limited number of practices (Ashikaga et al., 2002; Ayers & Kronenfeld, 2010), testing the latent structure of a larger and more comprehensive pool of TCAM practices may yield a different factorial solution. This should be additionally informative given that psychological profiles of consumers of different TCAM practices might differ.

Initial studies which explored the correlates of TCAM attitudes and use focused mostly on demographic and health-related factors. It has been repeatedly shown that TCAM users are more likely to be female, young (although the relationship with age may be more complex), more educated, and people of higher socio-economic status (for a review of early studies see Conboy et al., 2005), who perceive their health as poor and report having chronic pain (Palinkas & Kabongo, 2000) or chronic disease (Al-Windi, 2004). The main reasons for turning towards TCAM encompass failures of conventional treatments to cure a problem (Moore et al., 1985), the desire to avoid their adverse effects, and problems in communication with physicians (Vincent & Furnham, 1996). In the mid-2000s, research interest shifted toward psychological factors, yielding findings that TCAM attitudes and use can be predicted by various irrational beliefs and thinking patterns, including magical health beliefs (Aarnio & Lindeman, 2004), superstitions (Lourenco et al., 2008), paranormal beliefs (Lindeman, 2011), medical conspiracies (Fournier & Varet, 2023; Oliver & Wood, 2014) but also general conspiracist beliefs (Galliford & Furnham, 2017; Lamberty & Imhoff, 2018), as well as susceptibility to naturalness bias (O’Callaghan & Jordan, 2003), and cognitive biases related to probabilistic reasoning (Šrol, 2022; Teovanović et al., 2021). We use the term irrational mindset to refer to common features of these beliefs, i.e., that they do not conform to standards of normative logic, lack evidence-base and are persistent when confronted with disconfirming evidence (Lazarević et al., 2023; Žeželj & Lazarević, 2019).

Although there are several existing instruments for measuring TCAM, they typically either focus on attitudes and not on TCAM use (Bryden et al., 2018; Lindeman, 2011; Lobato et al., 2014) or measure specific TCAM practices (Lourenco et al., 2008; Van Den Bulck & Custers, 2010), relying on ad hoc selections of practices (Oliver & Wood, 2014) or even a single item (e.g. Peltzer & Pengpid, 2018), limiting their ability to provide a comprehensive assessment and examine patterns of actual TCAM use. To overcome these limitations and to address questions of empirical and psychological relevance of conceptual TCAM classifications, we aimed to develop a new, comprehensive measure of TCAM use. To capture the whole range of TCAM practices, we combined a top-down (literature search for existing TCAM taxonomies) and bottom-up approach (input from TCAM practitioners and content analysis of media reports). This allowed us to explore whether different categories of TCAM practices indeed group as the conceptual accounts propose and to test whether there is a latent proclivity to use all TCAM practices. To assess the relationship between TCAM use and other questionable health behaviors, we tracked intentional non-adherence to medical recommendations (iNAR, Purić, Petrović, et al., 2023). In this research, we formed a set of irrational beliefs with solid support of their predictiveness for TCAM use or attitudes. Assessing them in a single design made it possible to explore their mutual relations and, more importantly, compare their predictiveness for TCAM use. Measuring behaviors along with attitudes enabled us to compare the predictiveness of irrational beliefs for both.

The goals of the study were to (a) develop a comprehensive instrument on TCAM use; (b) explore its factorial structure; (c) relate the use of TCAM to iNAR; (d) track the roots of TCAM use, i.e., test whether they are best predicted by sociodemographics, health-related variables, ideological beliefs or a set of irrational beliefs; (e) explore the predictiveness of different irrational beliefs for TCAM use; (f) compare the predictive power of an identical set of variables for TCAM attitudes and TCAM use as outcomes.

Drawing from previous research, we preregistered the following hypotheses (outlined in Figure 1): iNAR and TCAM use will correlate weekly to moderately (Lamberty & Imhoff, 2018; Stanković et al., 2022) (H1). Use of TCAM will positively relate to superstition (Lindeman & Saher, 2007; Lourenco et al., 2008) (H2a), magical beliefs about health (Aarnio & Lindeman, 2004; Bryden et al., 2018; Lindeman, 2011) (H2b), conspiracy beliefs (Lamberty & Imhoff, 2018; Oliver & Wood, 2014; Stanković et al., 2022; Teovanović et al., 2021) (H2c) and attitude towards TCAM efficacy (Furnham, 2007) (H2d). TCAM use will be more common in women (H3) (e.g., Alwhaibi & Sambamoorthi, 2016; Barnes et al., 2004, 2008; Conboy et al., 2005; Stanković et al., 2022; Zhang et al., 2015), and will be negatively related to self-reported health status (Al-Windi, 2004) (H4a) and positively to the presence of chronic disease (Al-Windi, 2004; Barnes et al., 2008) (H4b). In the exploratory portion of the analyses, we related TCAM to illusory correlation (Smedslund, 1963), omission bias (Ritov & Baron, 1990), and naturalness bias (Meier & Lappas, 2016), as part of irrational mindset, as well as to religiousness and political orientation.

The design of the study and confirmatory analyses were preregistered (https://osf.io/pnugm). Two minor deviations from the preregistration are described at https://osf.io/62ezg. All materials, data¹, and analytic code required to reproduce the results of this study are publicly available at https://osf.io/9ktdg/.

To generate a comprehensive list of TCAM practices, we performed an extensive literature search in ScienceDirect, Medline, SAGE, Wiley, Cambridge University Press, Oxford Journals, Springer Link, and Google Scholar databases using keywords: alternative medicine, complementary medicine, CAM, traditional medicine, parochial medicine, folk medicine, herbal medicine, herbal remedies, unconventional/non-conventional medicine. Based on the results of this search, we identified an initial list of TCAM practices, which we supplemented with locally relevant practices based on the results of a focus group with TCAM practitioners (Purić et al., 2022) and the content analysis of online media articles on TCAM (Lazić et al., 2023).

Drawing from these sources, we came up with a list of 449 TCAM practices. We then removed duplicates, merged specific modalities of TCAM practices into larger groups, removed some which were deemed irrelevant to the local context, and added others that were relevant, coming up with a refined list of 88 practices. After removing remaining overlapping ones, we were left with the final list of 71 practices which were used as study items.

A sample size of N = 300 would allow us to perform factor analysis on the data (3-10:1 participant-to-variable ratio rule of thumb). At least N = 475 participants were required to detect at least two incidences of rare events with 1% frequency, with a power of 95% and alpha = .05.

Data were collected online using the snowball method and social networks. Respondents took part in the study on a voluntary basis and were not compensated for their participation. Inclusion criteria were Serbian residency and over 18 years of age. Data collection was terminated when the desired optimal sample size was reached (N = 475), and no new entries appeared in the database for five consecutive days. This resulted in an initial sample size of N = 646. After excluding participants who failed one or more out of three attention checks, the final sample of N = 583 (M_age = 39.01 years, SD_age = 12.10; 74.4% females) participants remained. All participants provided written informed consent. Data were collected in accordance with the Declaration of Helsinki, and the research was approved by the Psychology Research Ethics Committee at the Faculty of Philosophy, University of Belgrade, Serbia, reference number 935/1 (https://osf.io/bv7yh).

Unless otherwise indicated, responses to items were provided on a five-point Likert scale, from 1 - completely disagree to 5 - completely agree.

TCAM practices included 71 practices comprising: alternative medical systems, energy medicine, manipulative and body-based therapies, mind-body therapies, herbal medicine, biologically based treatments, traditional Serbian medicine, and COVID-19-specific pseudoscientific practices (for a list of items with initial categories see Supplement 1). For each TCAM practice, participants were asked to report if and when they used it (options: never heard about it/never used it/more than a year ago/in the past year/during the past two weeks). In addition, they were asked to recall their last use of each practice and report whether they used it for preventive purposes (advancing health), along with official therapy (complementary use), or instead of it (alternative use).

Intentional non-adherence to medical recommendations (iNAR, Purić, Petrović, et al., 2023) is an instrument consisting of 12 behavioral items which assess if and when participants chose not to adhere to medical recommendations (e.g., did not take prescribed therapy, took antibiotics without prescription).

COVID-19-specific behaviors were measured using 1) the number of COVID-19 vaccine doses participants received and 2) the extent to which they followed official health guidelines (e.g., mask-wearing, physical distancing, hand hygiene).

Health status was self-reported on a five-point scale (1 - very poor, 5 - very good). Participants also reported the presence of any chronic/long-term disease/condition using a checklist (e.g., cardiovascular, neurological, gastrointestinal). As a more objective, though imperfect health index (Gutin, 2018), we calculated the Body mass index (BMI) based on participants’ self-reported weight and height.

Attitude toward TCAM was measured using four items (e.g., I think that traditional, complementary, and alternative medicine is efficient).

Belief in conspiracy theories was assessed with the Conspiracy mentality questionnaire (Bruder et al., 2013, Serbian version by Lukić et al., 2019; Milošević Đorđević et al., 2021), consisting of five items (e.g., Many important things happen in the world which the public is never informed about).

Superstition was measured using five items from the Superstition scale (Žeželj et al., 2009) with the highest loadings on the general factor (e.g., I never walk underneath a ladder, even if it means I need to walk a longer distance).

Magical beliefs about health were assessed using three items from the Magical Beliefs About Food and Health scale (Lindeman et al., 2000), loading highest on the first factor (e.g., An incorrect diet makes food rot in the body).

Illusory correlation was assessed by presenting participants with a 2x2 contingency table showing no correlation between the appearance of knee pain and weather changes (Redelmeier & Tversky, 1996). Responses indicating either a positive or a negative correlation between these two variables were treated as indicating the presence of illusory correlation.

Omission bias was measured by presenting participants with a hypothetical scenario of having a medical condition connected with a 10% probability of developing chronic tooth pain (DiBonaventura & Chapman, 2008). They were asked to choose whether they would accept or decline a medication that cures this condition but brings a 5% probability of developing chronic tooth pain of the same intensity (four-point scale; 1 - would definitely decline medicine, 4 - would definitely accept medicine). Responses were recoded so that higher scores were indicative of a more pronounced omission bias.

Naturalness bias was assessed as a hypothetical preference for a natural drug over a synthetic drug, all other things being equal, in a situation that requires a choice between the two to treat an unspecified medical issue (Meier & Lappas, 2016).

We measured Religiousness with a single item (I consider myself a religious person).

Political orientation was assessed by a two-item measure of self-placement on the social and economic political axes using a 7-point scale (1 – extremely left-wing; 4 – center; 7 – extremely right-wing).

Sociodemographic block included gender, age, number of years of education, and socioeconomic status (SES) as a self-assessment of the monthly household budget (from 1 - We struggle to cover basic food expenses to 6 - Money is never an issue in our household).

Since we were interested in lifetime use of TCAM, as per preregistration, all TCAM items were binarized to reflect whether participants have ever used a given practice. This also increased the variability of TCAM practices, thus reducing the number of zero-frequency cells when calculating tetrachoric correlations. Binarized items were used to examine the latent composition of TCAM use. iNAR items, the number of COVID-19 vaccine doses, and the presence of chronic diseases/conditions were also binarized. All transformations and analyses were performed in R (R-4.2.0; R Core Team, 2022).

Since answers to TCAM and iNAR questionnaires were mandatory, there were no missing data for them (N = 583). For other instruments, the number of participants decreased due to survey dropout, resulting in N = 513 participants reaching the last question. Whenever possible, we deleted cases in a pairwise manner, except for regression analyses where, to enable model comparisons, all models were tested on a subset of N = 500 participants left after listwise deletion.

Herbal medicine practices (e.g., herbal teas, consumption of herbs, using herbal balms and ointments) as well as other biologically based treatments (e.g., vitamins/minerals/antioxidants and probiotics not recommended by a physician) were the most prevalent (> 70% each) (Supplement 1). Among mind-body therapies, practicing yoga (47%) and mindfulness/progressive muscle relaxation/breathing exercises (44%) were the most frequent ones. Pledges from brandy/cabbage/potato (72%) and using mineral/healing springs (41%) or holy water (31%) were the most frequently employed practices of traditional medicine, while visiting chiropractic/osteopathy/bonesetter practitioners (29%) or undergoing special massage treatments (28%) were amongst the most frequent body-based therapies. The use of alternative medical systems was rare in comparison – only homeopathy and aromatherapy/essential oil therapy were used by more than 15% of participants. Slightly over 10% of participants had an experience with energy medicine, such as (bio)energetic therapies and (electro)magnetic therapies. Finally, COVID-19-specific TCAM practices were observed in less than 10% of participants.

To explore the latent structure of TCAM use, we initially performed a minimum residual exploratory factor analysis with oblimin rotation on the tetrachoric correlation matrix of the full set of 71 TCAM items. Most TCAM practices showed loadings above .40 on the first factor (Supplement 2). However, as 22 practices were very infrequent (below 5%), there were numerous zero-frequency cells when estimating the tetrachoric correlations. Since the results of factor analysis on such a correlation matrix would be unreliable, we excluded these practices from further analyses. The Scree plot and Velicer’s minimum average partial (MAP) for the remaining 49 practices suggested a four-factor solution (see Supplement 2 for loadings). The TCAM practices with the highest loadings on the first latent factor, which we named Natural/biological products and practices, reflected using products such as extracts and supplements of both herbal and non-herbal origin, while the second factor, named Rituals/Customs, reflected the use of traditional medicine and religion-saturated practices. Mind-body therapies and energy medicine practices loaded on the third factor, New Age medicine, while alternative medical systems, certain energy medicines as well as body-based practices loaded on the fourth - Alternative medical systems.

To shorten the instrument while preserving its comprehensiveness, we merged some items based on the conceptual and correlational analyses, as well as their loadings on respective factors (e.g., praying and visiting monasteries for health). We further excluded 20 items with loadings below .50 on their respective factors. This procedure led us to the final selection of 22 items for measuring TCAM use (TCAM-22). The full step-by-step process, including a complete list of merged and excluded items, is available at https://osf.io/2taq7/.

The scree plot for the final set of 22 TCAM practices suggested a four-factor solution. Each practice significantly loaded on its respective factor, with practically no cross-loadings (Table 1). Additionally, almost all items loaded highly on the overall factor and the four factors were moderately correlated, so we performed a higher-order exploratory factor analysis. The scree plot, Parallel analysis, and Velicer’s MAP all pointed to a one-factor solution, and first-order factor loadings ranged from .46 (for Rituals/Customs) to .59 (for New Age medicine), justifying the calculation of an overall TCAM score.

Each factor of TCAM use, as well as the overall score, showed good internal consistency, as indicated by alpha coefficients (Table 1). The omega for the total scale was .80. TCAM-22 score correlated highly with the overall score of the initial 71-item set (r = .93, p < .001), showing that the reduction in the number of items was warranted.

On average, participants reported using around 40% of the 22 TCAM practices (Table 2); they mostly used them preventively (62%), while complementary (31%) and especially alternative (6%) uses were less present. All multi-item instruments demonstrated good internal consistency. The values of Skewness and Kurtosis were within the acceptable range of ±1 (Bulmer, 1979), although some distributions (mostly for one-item measures) deviated from normality to a larger extent.

In line with our predictions (H1), we found a positive correlation between TCAM use and intentional non-adherence to official recommendations (r = .27, p < .001). Those who tended to use TCAM more were also somewhat less likely to be vaccinated against COVID-19 (r = -.11, p = .009), although there was no correlation between TCAM use and adherence to recommended COVID-19 preventive behaviors (r = -.06, p = .18).

As expected, TCAM use was positively related to superstition (H2a), magical beliefs about health (H2b), and conspiracy mentality (H2c). It was positively related to all three cognitive biases and more present in women (H3). Contrary to expectations, TCAM use was not related to either self-reported health status (H4a) or the presence of chronic disease (H4b). We found only a moderate correlation between TCAM attitudes and TCAM use (r = .52, p < .001, H2d), supporting our decision to assess them separately. The full correlation matrix for all variables is in Supplement 3.

To further explore their predictors, we ran six hierarchical regression analyses with TCAM attitudes and TCAM use (overall and the four domains) as outcomes² (Table 3). In each of the analyses, we included sociodemographic variables in the first step, ideological beliefs in the second, indicators of health status in the third, and irrational mindset in the final, fourth step. The four blocks of predictors explained a total of 16.8% of the variance in TCAM use and 50.0% of the variance in TCAM attitudes. All blocks, with the exception of the health-related one, contributed significantly to TCAM use and TCAM attitudes. Religiousness, magical health beliefs, and naturalness bias significantly contributed to the prediction of both outcomes. Female gender contributed to the prediction of TCAM use but not attitude, while low SES and high conspiracy mentality contributed to the prediction of TCAM attitudes but not use.

With the four TCAM use domains as the outcomes, the predictors were somewhat different. Ideological beliefs played a substantial role only in Rituals/Customs. This was also the only domain where superstition - and no other irrational belief - played a critical role in the prediction. Younger age and religiosity also contributed to the prediction of Rituals/Customs. New age medicine domain was predicted by female gender, younger age, higher educational level, and magical beliefs about health. The Natural/biological products and practices domain was the least predicted one by our predictor set and was only predicted by magical beliefs about health and naturalness bias. The same two variables, together with female gender and older age, predicted the Alternative medical systems domain.

Drawing from the actual patterns of TCAM use in the Serbian population, factor analysis suggested that TCAM practices can be categorized into four unique domains: Natural/biological products and practices, Rituals/Customs, New Age medicine, and Alternative medical systems. This pattern aligns well with some often-used conceptual and/or practical taxonomies e.g., (Kaptchuk & Eisenberg, 2001; Wieland et al., 2011). The practices comprising Natural/biological products and practices closely correspond to the ones previously classified in this group: consuming herbal and honey-based products as well as taking vitamins and supplements. The Rituals/Customs domain includes praying and visiting monasteries, but also practices with parochial or folk elements such as wearing talismans, lead pouring, and using healing water. The grouping of practices in this domain supports a previous conclusion that praying is conceptually different from other mind-body interventions (Ayers & Kronenfeld, 2010) and suggests it is closer to traditional medicine. New Age medicine incorporates the bulk of what is typically classified as mind-body medicine (e.g., meditation, relaxation techniques, spiritual healing, mind-body exercises, visualization, expressive therapies). Finally, the Alternative medical systems domain encompasses the expected - acupuncture, homeopathy, and quantum medicine, but also (bio)energetic therapies and chiropractic.

We observed slightly different sociodemographic and psychological profiles of consumers of different TCAM practices. The well-established finding that women are more frequent users of TCAM (e.g., Barnes et al., 2004, 2008; Stanković et al., 2022; Zhang et al., 2015) held true for Alternative medical systems and New Age medicine but not for the other two domains. The role of age in TCAM use has previously been found to be inconsistent (Conboy et al., 2005), which may be because different studies assessed different TCAM practices. We found that younger people more often relied on New Age medicine and Rituals/Customs, older people on Alternative medical systems, while age played no role in the use of Natural/biological products and practices. The observed age-related pattern might be because the New Age domain encompasses recently developed or popularized practices, while the Rituals/Customs practices, although traditional, were only recently massively revived by media (Lazić et al., 2023). Participants who reported more use of Rituals/Customs practices were also (in terms of their ideological orientation) more religious, with stronger socially conservative views, as well as more superstitious; religiousness and superstitiousness did not predict the use of any other TCAM practices, further corroborating the distinctiveness of this group of practices. The use of Natural/biological products and practices was also distinct in that it was the least successfully predicted TCAM use domain. This is partially attributable to the fact that this domain was broadly defined, so as to encompass a number of common practices, causing a ceiling effect (56% of respondents obtained a maximum score).

When it comes to health-related predictors, overall TCAM use was associated only with a lower BMI, albeit weakly. Neither self-reported health status nor chronic illnesses were significant predictors, in contrast to previous studies (Al-Windi, 2004; Barnes et al., 2008). This is most likely because our participants most often practiced TCAM for preventive purposes (62% of responses) and not for treating existing medical conditions.

Beliefs and biases constituting irrational mindset incrementally predicted both TCAM attitudes and TCAM use (overall and the four domains) over and above sociodemographics, ideological beliefs, and reported health status. Attitudes are more likely to be better predicted by psychological factors, such as beliefs, than behaviors. This is because behaviors can be triggered by other cues like context, lack of willpower or resources. While previous studies have only demonstrated the importance of different aspects of irrational mindset for understanding TCAM attitudes (e.g., Jeswani & Furnham, 2010; Lindeman, 2011; Lourenzo et al., 2008), our results go one step further by testing the predictive power of a set of irrational beliefs in a single design and showing its central role in predicting not only TCAM attitudes but TCAM behaviors as well.

Two aspects of irrational mindset - magical health beliefs and naturalness bias - emerged as the strongest and most consistent predictors of these outcome variables. We believe this is due to their conceptual closeness (for discussion see Žeželj et al., 2023), but also to the fact that media reports and advertisements for TCAM practices often promote or rely on such magical ideas about health (e.g., that our body accumulates toxins and needs to be “detoxified”) and appeal to naturalness (e.g., that a product/practice is “nature’s gift” or “hidden secret of nature”) (Ernst, 2019; Lazić et al., 2023). Notably, a conceptually more distant manifestation of irrational mindset - conspiracy mentality - incrementally predicted TCAM attitudes, but not TCAM use. This is in line with previous research (Lamberty & Imhoff, 2018; Lobato et al., 2014; Teovanović et al., 2021), which suggests they might share a feature - a generalized distrust of official authorities, be it the dismissal of mainstream versions of controversial events or the dismissal of official medical recommendations and resorting to its alternatives. Indeed, we found that intentional non-adherence (e.g., self-medicating, skipping doctor’s appointments, stopping the course of antibiotics) was positively related to overall TCAM use and use of practices from all four domains, supporting previous research (Krousel-Wood et al., 2010; Owen-Smith et al., 2007). Similarly, and in line with other studies in the pandemic context (e.g., Teovanović et al., 2021), we found weak negative relations between following official COVID-19-related preventive behaviors and TCAM use.

Although they correlated with TCAM use in an expected manner, illusory correlation and omission bias lost their predictive power in the comprehensive predictive model. This could be because there is no content overlap between these biases and TCAM use. Biases seem to represent a basic, but remote mechanism underlying specific irrational beliefs including superstitions (Rudski, 2004), paranormal (Blanco et al., 2015), and pseudoscientific beliefs (Torres et al., 2020).

As one of our goals was to develop a novel measure of TCAM use, this study was conducted on a convenience sample. A study on a probability, nationally representative sample is needed to obtain information on the population prevalence of TCAM use.

We aimed for the general population and not for specific clinical groups (e.g., oncological patients, patients treated for infertility), who might have different TCAM use habits. The pattern of predictors might also be different for them (e.g. health-related variables might prove to be more important).

It can be argued that the four TCAM domains differ in how much they potentially harm (or benefit) someone’s health. This stems from their differing levels of evidence base and health risk, but also from the fact that they can be used with different goals in mind. Participants in our study reported using most of the TCAM practices preventively (62%), that is, for maintaining good health. However, when broken down per domain, different patterns of use emerge (for a detailed report, see Purić, Opačić, et al., 2023). Alternative medical systems were more frequently used in an alternative manner compared to other practices; replacing official treatments with treatments lacking strong evidence for their safety and efficiency (Ernst, 2019) may have the potential to cause the greatest harm. Natural/biological products and practices were more frequently used in parallel with official treatments; while complementary use is a less extreme choice, it runs the risk of interactions. Finally, New age medicine and Rituals/Customs were used almost exclusively for preventive purposes; while these practices may not be particularly effective in preventing disease, they are usually not harmful to health either. Future studies could account for these features to uncover any potential patterns in the associations between TCAM use and an irrational mindset. For example, scientifically backed health benefits – as well as the perceived ease-of-use – of some foods and vitamins (Natural/biological products and practices) may have downplayed the overall role of irrational beliefs; in contrast, some practices such as lead pouring (Rituals/Customs) are neither scientifically backed nor easily accessible and could therefore reflect a deeper commitment to a certain irrational belief.

These differences between domains are further reflected in their differential relations to intentional non-adherence - the strongest correlation was between the use of Natural/biological products and practices and iNAR (r = .27), while the other three domains showed lower correlations (from .13 to .17). Non-adherence might be more closely followed by use of Natural/biological products and practices as they are the most accessible and convenient to use; however, this should be explored in more depth in future studies.

Although we designed this study to incorporate a wide range of irrational beliefs and biases with a reliable evidence base for its relation with TCAM use, there are other potential candidates to be added to this group. Some could be content-specific, such as irrational health beliefs (Christensen et al., 1999) or medical conspiracy theories (Oliver & Wood, 2014); others could “go deeper” and include unexplored cognitive biases, such as belief bias (Evans et al., 1983), teleological bias (Kelemen & Rosset, 2009) and probabilistic reasoning biases such as base rate neglect, conjunction fallacy and gambler’s fallacy (Šrol, 2022; Teovanović et al., 2021).

We relied on retrospective self-report for TCAM use, which could have been biased (Schwarz, 2007). Future research could use a multi-method design and capture this behavior both retrospectively and via ambulatory assessment, such as experience sampling.

The correlational nature of this study did not allow for causal interpretations of the observed relations; experimental manipulation of irrational beliefs would bring evidence for the proposed causal chain. Furthermore, a chain with TCAM as the end behavioral point could be further traced down to broad cognitive styles like intuitive thinking (Wheeler & Hyland, 2008), and beyond - to illusory pattern perception (van Prooijen et al., 2018), apophenic tendencies, and proneness to psychotic-like behavior (Lazarević et al., 2021) - very basic dispositional mechanisms that are not only cognitive, but at the same time motivational.

Finally, the practices included in TCAM-22 vary regarding their cultural specificity. Whilst some of them are widespread (e.g., homeopathy, acupuncture, supplements, herbal remedies), others are more culturally specific (e.g., lead pouring). Thus, further cross-cultural validation of the factor structure of TCAM-22 is needed. In that respect, the observed relations between TCAM domains and irrational beliefs might vary across different cultural contexts. Therefore, further research is needed to demonstrate cross-cultural stability of the particular irrational beliefs as predictors of specific TCAM domains.

Our findings suggest that TCAM use cannot be explained by relying solely on sociodemographic variables or health status. Both attitudes toward TCAM and its actual use appear to be integrated with wider beliefs, most consistently to an irrational mindset. These findings partly correct and supplement existing models of health behavior that are based predominantly on rational factors (e.g., Abraham & Sheeran, 2015; Bosnjak et al., 2020; Schwarzer, 2008), suggesting that the inclusion of irrational beliefs would help better predict and understand certain questionable health behaviors such as TCAM use, although they may not be as important for others, such as intentional non-adherence to official medical recommendations (Purić, Petrović, et al., 2023). More broadly, our findings suggest that solely knowledge-based interventions (Sturgis & Allum, 2004) are not likely to be particularly effective. Instead of just providing correct information to consumers, interventions should seek to challenge the underlying (irrational) beliefs. Even though this might not be an easy goal, several types of interventions have previously proven effective with similar beliefs, e.g., cognitive debiasing (Ludolph & Schulz, 2018); myth debunking interventions, e.g., the myth of immoral “Big Pharma” and the moral CAM industry (Lamberty & Imhoff, 2018; Mijatović et al., 2022) inoculation (Jolley & Douglas, 2017) or narrative interventions against anti-vaccine conspiracy theories (Lazić & Žeželj, 2021), as well as exaggerated attitude-consistent messages (Hameiri et al., 2019).

Using a novel comprehensive instrument, we derived an empirical typology from the pattern of TCAM use, which corresponded closely to existing conceptual typologies, as well as to those based on attitudes toward TCAM, indicating that consumers perceive certain overarching characteristics of different TCAM treatments. Moreover, albeit similar, profiles of consumers using different TCAM domains showed important sociodemographic and psychological differences.

Nevertheless, what was common across all four domains of TCAM use, as well as for TCAM attitudes, was a ubiquitous relationship with a range of irrational mindset constructs - we suggest how this finding could inform public health interventions.

CRediT author statement: Conceptualization: all authors; Methodology: all authors; Investigation: all authors; Data curation: DP, MP, PL, MN, IŽ; Formal analysis: DP, MŽ, MP, PL, GK, PT, AL, GO, MB, LL, IŽ; Writing - Original Draft: all authors; Writing - Review and Editing: DP, MŽ, MP, PL, GK, PT, MN, AL, MB, LL, IŽ; Funding Acquisition: IŽ; Project Administration: DP, MP, IŽ; Supervision: DP, IŽ.

We would like to thank all medical experts and TCAM practitioners who took part in the focus groups and provided us with feedback on the versions of the instrument, as well as all respondents who took part in the online survey.

This research was supported by the Science Fund of the Republic of Serbia, #GRANT 7739597, Irrational mindset as a conceptual bridge from psychological dispositions to questionable health practices – REASON4HEALTH

The authors declare no competing interests.

The data and R code, along with all study materials are publicly available on the project OSF page: https://osf.io/9ktdg/.