Burning Mouth Syndrome Underlying Factors: A Roadmap From a Network Perspective

ABSTRACT

Objective

To investigate the relationship between biological, psychological, and social factors underlying Burning Mouth Syndrome (BMS).

Subjects and Methods

A case (n = 40) and control (n = 42) study containing 80 variables was examined using two network models based on regularized partial correlations (n = 82).

Results

The structure of the associative pathways with the BMS was revealed. Direct associations involved Gastrointestinal Alterations (0.23), Vitamin D Deficiency (0.29), Musculoskeletal Alterations (0.29), Symptom Severity Score 2 (SSS2) (0.22), Cortisol Variation (0.10), Interpersonal Sensitivity (0.04), Hostility (0.03). Global Severity Index, Symptom Severity Score 1, Psychoticism, Obsession‐Compulsion, Depression, Anxiety, and Somatization were indirectly related. The SSS2 was the most influential on BMS accuracy.

Conclusions

Gastrointestinal alterations and vitamin D deficiency show a significant influence on BMS while cortisol mediates in multiple associative pathways between musculoskeletal alterations, gastrointestinal alterations, vitamin D deficiency, non‐restorative sleep, fatigue, and cognitive problems. In addition to anxiety and depression, psychoticism, interpersonal sensitivity, and hostility stand out as psychological factors that seem to be related to a lack of vitamin D. None of the factors studied seem to have a relevant predictive potential for BMS, except for nonspecific symptoms of central sensitization.

1. Introduction

Burning Mouth Syndrome (BMS) is a persistent oral pain condition defined by the International Headache Society as “a recurring daily intraoral burning or dysaesthetic sensation lasting more than two hours per day for over three months, without apparent causative lesions” (Arnold 2018), and is considered an idiopathic pathology by the International Classification of Orofacial Pain (ICOP) (“International Classification of Orofacial Pain, 1st edition (ICOP)” 2020). Symptoms typically manifest on the tongue, lips, and hard palate and are often associated with altered taste perception (dysgeusia) and dry mouth (xerostomia). The precise underlying mechanisms of BMS remain uncertain. Currently, the pathogenesis of BMS is assumed to be a multifactorial process that encompasses central nervous system dysfunction, peripheral nerve neuropathy (Adamo and Spagnuolo 2022; Canfora et al. 2021), as well as inflammatory and psychological factors (Pereira et al. 2022). Notably, the ICOP categorizes BMS based on the presence or absence of somatosensory changes (“International Classification of Orofacial Pain, 1st edition (ICOP)” 2020). Some somatosensory changes, such as hypoalgesia, hyperalgesia, and allodynia, may indicate both central and peripheral neuropathy (Mignogna et al. 2011; Moisset et al. 2016; Schiavone et al. 2012a, 2012b).

So far, several biological factors potentially associated with BMS have been studied and the results have been inconclusive. In this sense, the role of local factors such as mechanical irritation, oral infections, allergic reactions, dentures, galvanism; or systemic factors such as gastrointestinal disorders, nutritional deficiencies (group B vitamins, zinc, and ferritin), urological or endocrine diseases, anemia, hyposialia and xerostomia, diabetes, etc. (Porporatti et al. 2023), remain controversial. Some of them, such as gastrointestinal and urogenital disorders, have even been proposed as risk factors for BMS (Netto et al. 2011). In addition, relationships between BMS, systemic diseases, and altered physiological analytical values have been described, which have generated controversy (de Pedro et al. 2020a; Fernandez‐Agra et al. 2023). Noteworthy, there is an observed deficiency of vitamin D in BMS patients (Morr Verenzuela et al. 2017). Although it has been suggested that primary BMS (idiopathic form of BMS) and secondary BMS (resulting from a local or systemic condition) be distinguished when any of these factors are present, some authors have argued that this remains a debatable issue (de Pedro et al. 2020b). The current understanding of BMS no longer distinguishes between primary and secondary forms. In contrast, it is now suggested that BMS should be viewed as a single condition with multiple potential contributing factors, rather than categorizing it based on the presence of an underlying condition (“Headache Classification Committee of the International Headache Society (IHS) The International Classification of Headache Disorders, 3rd edition,” 2018). This perspective aligns with the complexity of BMS, where symptoms can be severe even without an identifiable underlying medical condition. Finally, some studies have highlighted a complex interaction between local, systemic, psychogenic, and neuropathic factors involving BMS (Fernandez‐Agra et al. 2023; Lopez‐Jornet et al. 2020). The manner in which these factors are associated with BMS remains unknown.

BMS is a multifaceted disorder where psychological factors may play an essential role in its manifestation, progression, and impact on the individual's quality of life. It is widely accepted that BMS is associated with increased levels of anxiety, depression, and, to a lesser extent, neuroticism (Galli et al. 2017; Kim and Kho 2018). Furthermore, using the SCL‐90 evaluating nine types of psychopathology symptoms, BMS patients presented higher scores mainly on all scales compared to controls, that is for somatization, depression, anxiety, obsession‐compulsion, interpersonal sensitivity, depression, anxiety, hostility, and psychoticism (de Pedro et al. 2020a; Schiavone et al. 2012a, 2012b; Yoo et al. 2018). Keeping in mind the implication of psychological factors in BMS, various studies have assessed the salivary cortisol levels of patients with BMS observing that higher levels of anxiety are associated with higher salivary cortisol levels (Kim et al. 2012; Martínez et al. 2020). Furthermore, Kim, Kim, and Kho (2014) conducted a study with postmenopausal women, premenopausal women, and men diagnosed with BMS and found no differences between them in any of the nine symptomatic dimensions that this questionnaire evaluates (Kim, Kim, and Kho 2014), which would indicate that these parameters are very typical of BMS regardless of age or sex. Additionally, deficiencies in vitamin D have been linked to an elevated risk of experiencing depression and anxiety (Milaneschi et al. 2014). Nonetheless, the link between certain physiological factors, such as alterations in cortisol levels or Vit D deficiency, and the emotional state of patients with BMS remains unclear.

Altogether, the etiology of BMS remains elusive; a growing body of evidence suggests a significant interplay among multiple factors. However, the ways in which these factors relate to and interact with one another are far from well understood. To the best of the authors' knowledge, the combined influence of these factors has been poorly studied in patients with BMS. Furthermore, most studies assume the influence of various factors on BMS without considering the reverse effect. Consequently, the potential associative pathways and the predictive impact of these factors on BMS are not well understood. The main aim of this study was to investigate the complex relationship between biological, psychological, and social factors underlying BMS by using Network models.

Network models are useful for examining complex systems and the relationships between all the variables potentially involved without assuming them as predictors or “outcomes” a priori. These models have not been used in the BMS so far. Network models are advanced statistical methods that enable the analysis of the structure of relationships and the extent to which one variable can be predicted by others (Haslbeck and Waldorp 2020). In other words, it elucidates how the various components of the network are interconnected and to what extent they are influenced by external factors not accounted for within the network. This could contribute to a better understanding of the significance of relationships surrounding the BMS in practical contexts.

2. Methods

2.1. Participants

Data from 82 participants from a doctoral thesis (Monteserín Matesanz et al. 2020), which initially included 84 participants between March 2016 and February 2019, were utilized. The study protocol was approved by the San Carlos Hospital ethics committee (reference 2.2/16). The results of this study were partially published (Monteserin‐Matesanz et al. 2022). The research was conducted following the ethical principles of the “Declaration of Helsinki and good clinical practice guidelines,” and “the Strengthening the Reporting of Observational Studies in Epidemiology” (STROBE) guidelines. All participants were informed and gave their informed consent.

2.2. Study Design

A case–control design was used, which included a group of BMS patients (BMS group, n = 40) and a group of participants without BMS (Control group, n = 42). Patients with BMS received their diagnosis at the Department of Clinical Dental Specialties, which is part of the Faculty of Dentistry at the Complutense University of Madrid. Control group participants were selected from companions (relatives or friends of the patients), matched for age and sex with the BMS patient group, using a case–control ratio of 1 to 1. All participants were evaluated and diagnosed by a single expert clinician in oral medicine. The participants with BMS were new patients or patients experiencing a relapse.

The inclusion criteria in the BMS group were Diagnosis of BMS according to the diagnostic criteria of the third edition β of the International Classification of Headache Disorders (ICDH‐IIIβ) of the International Headache Society (IHS) (Headache Classification Committee of the International Headache 2013).

The inclusion criteria for the control group were the absence of clinical symptoms of BMS or any other pathology of the oral mucosa such as Sjögren's syndrome, candidiasis, recurrent aphthous stomatitis, oral lichen planus, subplaque stomatitis, exudative multiforme erythema, etc.

Both groups included individuals over 18 years of age who are legally competent. The exclusion criteria for both groups were a history of oncological treatment (radiotherapy and chemotherapy) in the last 5 years, uncontrolled systemic diseases, treatment with narcotic drugs or antiepileptics during the last 3 months, and difficulties in understanding language.

2.3. Measures

The same examiner recorded physical examination and objective salivary measures. Additionally, self‐reported measures were also collected. All measurements were recorded during a single visit after the diagnosis, except for the objective salivary measurements, which were recorded during a second visit.

The measure from the physical examination of the mouth included the presence of amalgam fillings or the use of dental prosthesis. The objective salivary measures were salivary flow and salivary biochemical parameters. For the biochemical parameters, two saliva samples were used. The first was collected between 9:00 and 10:00 a.m., and the second between 12:00 and 1:00 p.m. on the same day. For cortisol (nmol/L) and magnesium (mg/dL) measurements, participants were instructed not to eat or drink at least 1 h before the sample was taken. This sample was kept refrigerated between 6° and 8°C until it was sent to the laboratory where it was stored in ultra‐freezing (−80°C). Between the collection of both saliva samples, the resting salivary flow and the stimulated salivary flow were recorded by salivary drainage (Navazesh and Kumar 2008). More details are given in the Supporting Information. Regarding cortisol, the variable Cortisol Variation was constructed as the difference between the salivary cortisol levels in the second sample (12:00–1:00 p.m.) and the first (9:00–10:00 a.m.).

The self‐reported measures were as follows: gender, age, level of education, profession, current occupation, personal habits, nutritional deficits, menopause, perimenopause, systemic diseases, pharmacological treatment, intensity of symptoms (Visual Analog Scale [VAS], 0 to 10), duration of the condition (months), type of symptoms, number of symptoms, location of symptoms, number of locations of symptoms, type of BMS according to Lamey and Lewis (1989), variables related to saliva (taste alterations and subjective salivation sensation), variables related to psychological aspects, oral habits, triggers, and factors modifying the symptoms.

The self‐reported measures recorded using psychometric instruments were Symptom Checklist‐90‐Revised (SCL‐90R) scores and Symptom Severity Score (SSS). Therefore, seven subscales (range: 0–4) and the Global Severity Index (GSI; range: 0–4) of the revised and validated version in the Spanish population of the SCL‐90R (González de Rivera et al. 2002) were used, which evaluate a wide range of psychopathological symptoms and the intensity of psychic suffering, respectively (Derogatis, Lipman, and Covi 1973). GSI is a general measure of psychological distress, calculated by summing the responses to the 90 items of the SCL‐90R and averaging the obtained score. A higher GSI indicates greater symptom severity and suggests a higher intensity of psychological distress. The subscales used were Somatization (SOM), Obsession‐Compulsion (OBS), Interpersonal Sensitivity (INT), Depression (DEP), Anxiety (ANX), Hostility (HOS), Phobic Anxiety (PA), and Psychoticism (PSI). The SSS is an instrument composed of two scales: the first (SSS1) measures the non‐restorative sleep, fatigue, and cognitive problems; and the second (SSS2) assesses the number of nonspecific symptoms of central sensitization (range: 0–3). In both scales, the total score was obtained by adding the scores of each question, with higher scores indicating greater symptom intensity.

2.4. Data Analysis

The statistical software R (version 4.0.3) and its interface R Studio (R Studio Team) were used. To examine the structure of the associative pathways, as well as the direct and indirect effects on BMS, a preliminary selection of variables was first conducted from an initial set of 80 variables. The first selection criterion was retaining variables that showed significant differences between the BMS and control groups, with a threshold of α = 0.1. For this purpose, Student's t‐tests, the Wilcoxon rank‐sum exact test, Pearson's chi‐squared test (χ ²), Fisher's exact test, and generalized estimating equations (GEE) were used after verifying the goodness of fit to a normal distribution for the continuous variables. The second criterion was to remove variables that were either unconnected to any other variable or that limited or prevented the network model's convergence. The third criterion was conceptual: the use of anxiolytics and antidepressants, as well as other self‐reported psychological aspects, was excluded to avoid increasing the number of variables and complicating the interpretation of the results. Instead, the subscales of the SCL‐90R were retained. Additionally, BMS symptoms were excluded as they provided redundant information already encompassed in the BMS diagnosis. The complete list of selected and discarded variables is presented in the Supporting Information.

Subsequently, two network models (pairwise Mixed Graphical Models [MGMs]) were estimated using the selected variables. The first model, named “Somatic network,” incorporated seven variables along with BMS: Vit D Deficiency (Vit D), Gastrointestinal Alterations (Gastro), and Musculoskeletal Alterations (Muskel) as binomial variables (yes or no); Cortisol Variation, the total score of SSS1, and GSI as Gaussian variables; and the total score of SSS2 as a Poisson variable. The second, named “Psychosomatic network,” incorporated 10 variables in addition to BMS: Cortisol Variation, the seven components of the SCL‐90R (SOM, OBS, INT, DEP, ANX, HOS, and PSY), and SSS2 as Gaussian variables; and Vit D Deficiency as a binomial variable (yes or no). Cortisol Variation and Vit D deficiency were added to both networks due to their known relationship with physiological and psychic aspects. The stability of both networks was examined using Bootstrap sampling distributions. The MGMs support variables of different types and provide a solution that can be viewed as an undirected network of conditional dependence relationships (represented by the edges of the network) between the variables (represented by the network nodes). The estimated parameters are combined into an aggregated edge‐weight (network edges) and measure the strength of association between variables, allowing to distinguish between direct and indirect effects. These parameters were regularized to avoid overly complex solutions and reduce the risk of overfitting. This method provides network edge weights that are zero, so there is no need for significance tests or post hoc thresholds (Altenbuchinger et al. 2020). Cross‐validation techniques and regularization methods used allow for obtaining robust results even in small to moderate samples, particularly for mixed and multigroup data (Altenbuchinger et al. 2020; Meinshausen and Bühlmann 2006). A direct relationship was considered when a network edge connected two nodes, while an indirect relationship was established when two nodes were connected through at least one other node.

In this research, a node‐wise estimation approach was used. In other words, the model was estimated by taking each node in turn and performing a regression of all other nodes on it while keeping the other variables in the model. Therefore, for a pairwise interaction (A and B), two specific effects are provided: one from the regression on A (B → A) and another from the regression on B (A → B), correcting for all other variables present in the model (Haslbeck and Waldorp 2020). The details about the estimation are detailed in the Supporting Information. Finally, the stability of the edge weights (or network stability) was examined using bootstrap sampling distributions (random number seed “1234”; 70–100 bootstrapped sampling distributions). In other words, different samples were generated from the original dataset, and the edge weights were recalculated for each one, resulting in their sampling distributions. The variation in estimates across all samples was then observed.

To study the relevance of the remaining variables concerning BMS, predictability measures were computed and the effect of removing each variable on the predictability of BMS was also examined. For this purpose, the percentage of explained variance, the percentage of correct classification (accuracy), and the normalized accuracy (accuracy_norm) were used. The latter indicates how much a variable is predicted by the remaining variables in the network, beyond the trivial prediction by the marginal distribution (Haslbeck and Waldorp 2018). That is, it adjusts the accuracy considering the probability of random guessing. The visualization of the model and predictability was carried out through the “qgraph” package. In the design, the position of the nodes was achieved through the Fruchterman Reingold algorithm so that the edges have a similar length and overlaps are avoided.

3. Results

3.1. Participants

Data from 84 participants were used: 40 were patients with BMS and 44 were individuals without BMS. Two individuals without BMS were excluded because they refused to attend the second visit. Therefore, the final sample (N = 82) consisted of 40 patients in the BMS group and 42 individuals in the control group without BMS. No differences were observed in the sociodemographic characteristics and personal habits between the two groups (Table 1). The selected variables, including those related to the systemic status, saliva‐related variables, psychological aspects, other potentially BMS‐related factors, and scores from the SCL‐90R, SSS1, and SSS2, are presented in Table 2 and Tables S2.3–S2.8, respectively.

TABLE 1.

Sociodemographic characteristics and personal habits.

	BMS group (n = 40)	Control group (n = 42)	p
Gender, n (%)
Male	3 (7.5)	3 (7.1)	0.95
Female	37 (92.5)	39 (92.9)
Age (years), mean (SD), range	62.36 (13.3)	61.68 (9.1)	0.5
	27.88 to 81.9	30.63 to 82.1
Education level, n (%)			0.13
Incomplete primary education	5 (12.5)	4 (9.5)
Complete primary education	9 (22.5)	20 (47.6)
First stage primary education	4 (10)	2 (4.8)
Second stage primary education	7 (17.5)	2 (4.8)
Higher degree or vocational training	4 (10)	6 (14.3)
University studies	11 (27.5)	8 (19)
Occupation, n (%)			0.41
University degrees	10 (25)	2 (4.8)
University diplomas	1 (2.5)	7 (16.7)
Intermediate occupations and self‐employed individuals	7 (17.5)	10 (23.8)
Supervisors and skilled workers	3 (7.5)	3 (7.1)
Primary sector and semi‐skilled workers	10 (25)	8 (19)
Unskilled workers	9 (22.5)	12 (28.6)
Current occupation, n (%)			0.71
Employed	17 (42.5)	18 (42.9)
Unemployed	2 (5)	4 (9.5)
Retired/early retired	18 (45)	14 (33.3)
Unable to work	0	1 (2.4)
Homemaker	3 (7.5)	5 (11.9)
Personal lifestyle and health behaviors, n (%)
Tobacco consumption	9 (22.5)	7 (16.7)	0.78
Number of cigarettes per day (SD)	11.78 (9.9)	11.29 (4.7)	0.90
Alcohol consumption	21 (52.5)	28 (66.7)	0.19
Number of standard drinks per week (NSD)	7.33 (1.3)	5.75 (0.9)	0.31
Caffeine consumption	37 (92.5)	36 (85.7)	0.33
Daily milligrams of caffeine, mean (SD)	145.43 (155.1)	111.71 (122.1)	0.28
Alcohol mouthwash use	4 (10)	7 (16.7)	0.38
Hot foods	10 (25)	17 (40.5)	0.14
Spicy foods	4 (10)	5 (11.9)	0.78
Daily water intake
< 2 L	3 (7.5)	5 (11.9)	0.96
1.5–2 L	20 (50)	17 (40.5)
0–1.5 L	17 (42.5)	20 (47.6)

Note: NSD, standard drink units (containing between 8 and 13 g of alcohol per unit).

TABLE 2.

Systemic conditions.

	BMS group (n = 40)	Control group (n = 42)	p
Nutritional deficits, n (%)
Vitamin B12 deficiency	2 (5)	2 (4.8)	0.96
Iron deficiency	5 (12.5)	4 (9.5)	0.67
Vitamin B1, B2, B6 deficiencies	0	0	—
Vitamin D deficiency	8 (20)	3 (7.1)	0.09
Menopause, n (%)	30 (75)	36 (85.7)	0.22
Perimenopause, n (%)	3 (7.5)	1 (2.4)	0.29
Systemic diseases, n (%)
Hypertension	14 (35)	12 (28.6)	0.53
Diabetes mellitus
Type I	0	1 (2.4)
Type II	3 (7.5)	3 (7.1)	0.56
Thyroid disorders
Hypothyroidism	5 (12.5)	3 (7.1)
Subclinical hypothyroidism	1 (2.5)	0	0.25
Hyperthyroidism	1 (2.5)	0
Subclinical hyperthyroidism	0	1 (2.4)
Gastrointestinal disorders	16 (40)	5 (11.9)	0.004
Cardiac disorders	4 (10)	5 (11.9)	0.784
Pulmonary disorders	2 (5)	2 (4.8)	0.960
Cancer	5 (12.5)	2 (4.8)	0.213
Liver disorders	5 (12.5)	3 (7.1)	0.417
Renal disorders	4 (10)	2 (4.8)	0.366
Musculoskeletal disorders	3 (7.5)	10 (23.8)	0.043
Bone disorders	15 (37.5)	16 (38.1)	0.956
Fibromyalgia	2 (5)	1 (2.4)	0.530
Autoimmune disorders	4 (10)	3 (7.3)	0.669
Ocular disorders	4 (10)	6 (14.3)	0.556
Gastroesophageal reflux	15 (37.5)	7 (16.7)	0.033
Lipid metabolism disorders	9 (22.5)	20 (47.6)	0.017
Adrenal disorders	0	0	—
Headaches and migraines	3 (7.5)	1 (2.4)	0.285
Pharmacological Treatment, n (%)
Antihypertensives
ACE inhibitors	4 (10)	4 (9.5)	0.942
Beta‐blockers	5 (12.5)	1 (2.4)	0.080
ARBs (Angiotensin II Receptor Blockers)	6 (15)	5 (11.9)	0.681
Oral antidiabetic drugs	2 (5)	2 (4.8)	0.960
Insulin	1 (2.5)	1 (2.4)	0.972
Thyroid medication	9 (22.5)	4 (9.5)	0.108
Diuretics	11 (27.5)	6 (14.3)	0.140
Gastric protectors	16 (40)	9 (21.4)	0.068
Antiplatelet agents	5 (12.5)	5 (11.9)	0.935
Anticoagulants	0	0	—
Hypolipidemic agents	6 (15)	14 (333)	0.053
Nonsteroidal anti‐inflammatory drugs (NSAIDs)	7 (17.5)	9 (21.4)	0.654
Bisphosphonates	0	1 (2.4)	0.329
Calcium or Vitamin supplements	11 (27.5)	6 (14.3)	0.140
Anxiolytics	16 (40)	9 (21.4)	0.068
Antidepressants	13 (32.5)	2 (4.8)	0.001
Antihistamines	1 (2.5)	3 (7.1)	0.332
Alternative medicine	8 (20)	5 (11.9)	0.316

Note: In bold, those selected variables (p < 0.1).

3.2. Somatic Network

The network analysis revealed direct relationships between the BMS and five remaining variables: Vit D deficiency, Gastrointestinal Alterations, Musculoskeletal Alterations, Cortisol Variation, and SSS2. The indirect relationships between BMS‐GSI and BMS‐SSS1 were also identified. The structure of these associative pathways is shown in Figure 1.

FIGURE 1.

Structure of the associative pathways around the BMS in the Somatic Network. It is observed that Gastrointestinal Alterations, SSS2, and Cortisol Variation mediate the indirect relationship between BMS‐SSS1; and Vitamin D Deficiency and SSS2 mediate the BMS‐GSI relationship. The rest of the relationships around the BMS were direct. Cortisol Variation stood out as it has multiple connections and participates in the relationships between BMS–Vitamin D Deficiency, BMS–Musculoskeletal Alterations, BMS–Gastrointestinal Alterations, and BMS–SSS1. Green edges indicate positive relationships, red edges indicate negative relationships and gray edges indicate relationships with categorical variables. The width of the edge is proportional to the strength (absolute value) of the edge. The blue ring indicates the proportion of explained variance (for continuous nodes) and accuracy (for binary nodes). For the BMS node, the orange part of the ring shows the accuracy of the intercept model, while the red part of the ring is the “additional” accuracy achieved by all the remaining variables. The sum of both is the accuracy of the full model (A) and the normalized accuracy A_norm is the ratio between the additional accuracy due to the remaining variables (red) and one minus the accuracy of the intercept model (white + red). BMS, burning mouth syndrome; Cortisol, cortisol variation; Gastro, gastrointestinal alterations; GSI, psychic and psychosomatic suffering; Muskel, musculoskeletal alterations; SSS1, non−restorative sleep, fatigue and cognitive problems; SSS2, non−specific symptoms of central sensitization; Vit D, vitamin D deficiency.

By breaking down each of the connections with BMS, bidirectional relationships (effect of other variables on the BMS and vice versa) were observed with Gastrointestinal Alterations, Musculoskeletal Alterations, Vit D Deficiency, and SSS2. This showed that the probability of BMS increased with Gastrointestinal Alterations, Vit D deficiency, and SSS2. The effect on the BMS was greater than the opposite effect only in the BMS‐Gastrointestinal Alterations relationship. The BMS increased Cortisol Variation being the only unidirectional relationship (Table 3). All relationships were conditioned on keeping the other variables in the model constant.

TABLE 3.

Regression parameters using BMS as predictor and outcome.

	Somatic network			Psychosomatic network
	Aggregated edge‐weight	Regression on BMS	Regression using BMS as predictor	Aggregated edge‐weight	Regression on BMS	Regression using BMS as predictor
Gastrointestinal disorders	0.23	0.35	0.12	—	—	—
Musculoskeletal disorders	0.29	−0.29	−0.30	—	—	—
Vitamin D deficiency	0.29	0.24	0.34	0.09	0.19	0
SSS2	0.22	0.09	0.34	0.24	0.09	0.38
Cortisol variation	0.10	0	−0.19	0.19	−0.09	−0.29
SSS1	0	0	0	—	—	—
GSI (Global Severity Index)	0	0	0	—	—	—
Psychoticism (PSI)	—	—	—	0	0	0
Obsession‐compulsion (OBS)	—	—	—	0	0	0
Interpersonal sensitivity (INT)	—	—	—	0.04	0.09	0
Depression (DEP)	—	—	—	0	0	0
Anxiety (ANS)	—	—	—	0	0	0
Hostility (HOS)	—	—	—	0.03	0.05	0
Somatization (SOM)	—	—	—	0	0	0

The SSS2 was the most relevant variable regarding the predictability of the BMS, followed by Gastrointestinal Alterations, Musculoskeletal Alterations, and Cortisol Variation. The removal of these variables resulted in a reduction of the accuracy and accuracy_norm of the BMS. On the other hand, the remaining variables removal did not reduce the predictability of the BMS (Table 4).

TABLE 4.

Effect of removing variables on the predictability of BMS.

Removed variables	BMS accuracy	BMS accuracynorm
Somatic network
Symptom severity score 2 (SSS2)	0.70	0.38
Gastrointestinal disorders	0.74	0.48
Musculoskeletal disorders	0.76	0.50
Cortisol variation	0.77	0.53
Vitamin D deficiency	0.78	0.55
Symptom severity score 1 (SSS1)	0.78	0.55
Global severity index (GSI)	0.78	0.55
Psychosomatic network
Symptom severity score (SSS2)	0.70	0.38
Cortisol variation	0.73	0.45
Psychoticism (PSI)	0.74	0.48
Vitamin D deficiency	0.74	0.48
Obsession‐compulsion (OBS)	0.76	0.50
Interpersonal sensitivity (INT)	0.76	0.50
Depression (DEP)	0.76	0.50
Anxiety (ANS)	0.76	0.50
Hostility (HOS)	0.76	0.50
Somatization (SOM)	0.77	0.53

3.3. Psychosomatic Network

The network analysis revealed direct relationships between the BMS and five of the remaining 10 variables: SSS2, Cortisol Variation, Vit D Deficiency, HOS, and INT. BMS was indirectly related to the remaining components of the SCL90‐R (ANX, DEP, OBS, and PSY). The structure of these associative pathways is shown in Figure 2.

FIGURE 2.

Structure of the associative pathways around the BMS in the Psychosomatic Network. Three association pathways of the BMS with the components of the SCL90‐R (nodes with green ring) were identified: The first involved the indirect relationship BMS‐HOS through Cortisol Variation and Vitamin D Deficiency, and the direct relationship BMS‐HOS which was weaker; the second involved the indirect relationship BMS‐SOM through SSS2; while the third was the direct relationship BMS‐INT. Green edges indicate positive relationships, red edges indicate negative relationships, and gray edges indicate relationships with categorical variables. The width of the edge is proportional to the strength (absolute value) of the edge. The ring of each node indicates the proportion of explained variance (for continuous nodes) and accuracy (for binary nodes). For the BMS node, the orange part of the ring shows the accuracy of the intercept model, while the red part of the ring is the additional accuracy achieved by all the remaining variables. The sum of both is the accuracy of the full model (A) and the normalized accuracy A_norm is the ratio between the additional accuracy due to the remaining variables (red) and one minus the accuracy of the intercept model (white + red). ANS, anxiety; BMS, burning mouth syndrome; Cortisol, cortisol variation; DEP, depression; HOS, hostility; INT, interpersonal sensitivity; OBS, obsession−compulsion; PSI, psychoticism; SOM, somatization; SSS2, central sensitization symptoms; Vit D, vitamin D deficiency.

By breaking down each of the connections with BMS, bidirectional relationships were observed with SSS2 and Cortisol Variation, with the effect on BMS being lower. In addition, unidirectional relationships with Vit D Deficiency, INT, and HOS, on the BMS were identified. Vit D Deficiency, SSS2, Cortisol Variation, INT, and HOS increased the likelihood of BMS (Table 3).

The SSS2 was the most relevant variable on the predictability of the BMS followed by Cortisol Variation, PSY, and Vit D Deficiency. On the contrary, the others were not relevant since their removal did not reduce the predictability of the BMS (Table 4). The associations that involved the BMS were considered relatively stable since the proportion of estimates different from zero in the samples was > 0.5 (Figures S5 and S6).

4. Discussion

BMS is a complex and poorly understood painful condition. This cross‐sectional study shows a structural image of the associative pathways between some of the most relevant factors in BMS patients, which should be considered interaction patterns and not strictly causal. Additionally, the specific effects of these factors on the BMS are examined, and their limited ability to predict or classify individuals as BMS patients is revealed.

This study shows that some aspects are directly linked to the BMS (Gastrointestinal Alterations, nonspecific symptoms associated with central sensitization conditions, Vit D Deficiency, morning variation of cortisol levels, and musculoskeletal alterations), while others are also indirectly linked to the BMS (Non‐restorative Sleep, Fatigue and Cognitive problems and the intensity of psychic and psychosomatic suffering). In general, the associative pathways involving BMS are weak to moderate, with Gastrointestinal Alterations and Vit D Deficiency exerting the greatest influence on BMS. Each of these associative pathways is discussed below.

4.1. Gastrointestinal Alterations

The first associative pathway of the somatic network shows a bidirectional relationship in which the effect of Gastrointestinal Alterations on the BMS is greater than the opposite effect. In addition, Gastrointestinal Alterations appear to influence the BMS more than any other aspect examined (Table 3). Additionally, Gastrointestinal Alterations are shown as a mediator in the associations BMS–Non‐restorative Sleep, Fatigue and Cognitive problems, and BMS–Cortisol Variation (Figure 1).

So far, the available evidence on the potential association between Gastrointestinal Alterations and the BMS is limited. In line with our findings, BMS patients exhibit poorer health status and are more likely to have gastrointestinal diseases (OR = 3.8) (Netto et al. 2011) compared to controls (de Pedro et al. 2020a; Lamey et al. 2005). Even gastrointestinal problems have been pointed out, along with urogenital problems, as risk factors associated with BMS (Netto et al. 2011). Among the indicators of gastrointestinal problems, both the consumption of gastro‐protectors and the influence of gastroesophageal reflux in BMS patients are controversial (Azzi et al. 2019; Netto et al. 2011). The latter is one of the most common comorbidities associated with BMS and both conditions have been linked independently of reflux acidity (Lechien et al. 2021). In our study, both gastroesophageal reflux and the consumption of gastric protectors caused analytical problems, which cast doubt on their potential effect on the BMS. On the other hand, gastrointestinal alterations and BMS could be linked through alterations in pain modulation, or central sensitization, which are associated with overactivation of the hypothalamus–pituitary–adrenal axis (HPA axis) and the efferent sympathetic branch of the autonomic nervous system, which influences the digestive function (Gottfried‐Blackmore, Habtezion, and Nguyen 2021). This could be particularly relevant in some disorders of gut–brain interaction such as functional dyspepsia, bowel syndrome, functional constipation, or Crohn's disease, among others.

Recent findings support the role of gastrointestinal alterations as a mediator in two indirect associative pathways: BMS–Non‐restorative Sleep, Fatigue, and Cognitive problems, and BMS–Cortisol Variation. The first associative pathway shows a notable interaction between Gastrointestinal Alterations and Non‐restorative Sleep, Fatigue, and Cognitive problems; three aspects that should be interpreted together as they were analyzed in combination. Cognitive deterioration could be associated with alterations in the microbiome–gut–brain axis that can alter the permeability of the blood–brain barrier and generate neuroinflammation (Cattaneo et al. 2017; Liang et al. 2022). Recently, it has been pointed out that some intestinal microbial species might be related to cognitive deterioration (Liang et al. 2022), which is a characteristic aspect of patients with BMS (Canfora et al. 2021). Additionally, some metabolic alterations related to sleep deterioration may be mediated by the excess of specific intestinal bacteria. Particularly, both fragmentation and short duration of sleep have been associated with intestinal dysbiosis, possibly due to the HPA axis activity. Furthermore, bacterial species that proliferate because of sleep deterioration can produce fatigue (Matenchuk, Mandhane, and Kozyrskyj 2020). The second associative pathway is discussed in the subsection “Variation in cortisol levels.”

4.2. Cognitive Impairment, Fatigue, and Sleep Disturbances

This study shows that Non‐restorative sleep, Fatigue, and Cognitive problems mediate in the associations BMS–Psychic Suffering, BMS–Non‐specific Central Sensitization Symptoms, and BMS–Cortisol Variation. Their relationship with the BMS is established indirectly through the variation in cortisol levels, the central sensitization symptoms, and gastrointestinal alterations.

Although a causal relationship between sleep disturbances and BMS has not been demonstrated, these are more frequent in BMS patients than in controls (Alhendi et al. 2023; Leuci et al. 2023). Particularly, subjective sleep quality, sleep latency, and the use of sleeping medication seem to be the most affected components (Canfora et al. 2021). Also, shorter and poor‐quality sleep has been reported in 75% of BMS patients compared to 52% of healthy individuals (Canfora et al. 2022). On the other hand, patients with BMS exhibit a state called “burning fog,” linked to difficulties in concentration, forgetfulness, confusion, difficulty in performing multiple tasks, and a decrease in cognitive functions (working memory, attention, and executive functions) (Canfora et al. 2021). The exact mechanism linking sleep deterioration with cognitive function alterations is not clear. Nonetheless, periods of short or prolonged sleep, poor sleep quality, or insomnia have been associated with a high risk of cognitive deterioration (Shi et al. 2018). Furthermore, mood disorders and sleep disorders have been observed in both BMS patients and patients with cognitive impairment (Canfora et al. 2021; Chan et al. 2020). Therefore, it is hard to rule out the multiple factors that may affect cognitive functions and worsen pain perception in BMS patients (Canfora et al. 2021).

The association between non‐restorative sleep and psychological distress is reasonable (Canfora et al. 2022; Jaaskelainen and Woda 2017). However, the temporality and causal relationship between BMS and mood disorders are unclear. Thus, sleep disorders are also common in other chronic conditions such as depression, anxiety, hypertension, diabetes, and obesity (Adamo et al. 2018). In addition, sleep impairment has shown a weak association with the intensity and quality of pain in BMS patients (Adamo et al. 2018), which is consistent with the inhibition of pain during sleep, its progressive increase throughout the day, and its relief when eating and drinking (Canfora et al. 2022). Taken together, these findings suggest that psychological distress, gastrointestinal disturbances, central sensitization symptoms, and variation in cortisol levels, could be more influential in the sleep impairment of BMS patients than the specific characteristics of the pain. Consequently, this study supports that psychological distress and pain should be evaluated separately to better understand each characteristic of a patient affected by BMS (Canfora et al. 2022).

4.3. Nonspecific Symptoms of Central Sensitization

The second associative pathway directly links BMS and Nonspecific Central Sensitization Symptoms, revealing a greater effect of BMS on the central sensitization symptoms compared to the minimal influence of these symptoms on BMS (Table 3). This suggests that BMS enhances central sensitization and is indirectly affected by other conditions that also generate sensitization through their consequences on sleep, cognitive problems, and mood disorders. This central sensitization based on altered pain modulation could induce burning pain perception in BMS patients (Imamura et al. 2019). Consequently, the central sensitization symptoms (Yunus 2007) behave as a shared mediator in two associative pathways: BMS–Non‐restorative Sleep, Fatigue, and Cognitive problems and, BMS–Psychological Distress (Figure 1). This impact of BMS on the Nonspecific Central Sensitization Symptoms is consistent with the usual comorbidity of BMS with other conditions that generate central sensitization such as fibromyalgia, low back pain, chronic fatigue syndrome, or vulvodynia (Adamo and Spagnuolo 2022; Leuci et al. 2023). In parallel, some chronic gastrointestinal problems related to the activity of the gut–brain axis could serve as a relevant source of central sensitization (Gottfried‐Blackmore, Habtezion, and Nguyen 2021) in BMS patients. The central sensitization related to the overlap of these conditions could contribute to the pain modulation system being overused in the BMS brain (Imamura et al. 2019).

The influence of central sensitization symptoms on the relationship between BMS and sleep impairment is consistent with greater sleep problems in those individuals who exhibit central sensitization symptoms as in neuropathic pain (Aoyagi et al. 2022; Sachau et al. 2023). Similarly, altered endogenous pain modulation has been linked to sleep disturbances in other central sensitization conditions, such as migraine (Sachau et al. 2023). In addition, some indicators of central sensitization such as thermal hyperalgesia and burning have been associated with the adequacy and amount of sleep, but not with sleepiness, similar to sleep disturbances reported in BMS patients (Adamo et al. 2018). Although sensory abnormalities linked to central sensitization were not evaluated in this study, these interactions are consistent with the BMS–Central Sensitization Symptoms–Psychological Distress associative pathway.

4.4. Vitamin D Deficiency

The third associative pathway involving BMS reveals a bidirectional relationship with Vit D Deficiency. This aspect shows an effect on BMS slightly lower than the effect of BMS on Vit D Deficiency (Table 3). Similarly, Vit D Deficiency is revealed as a shared mediator in two associative pathways linked to BMS: BMS–Cortisol Variation and BMS–Psychological Distress (Figure 1).

The potential relationship between Vit D Deficiency and BMS is supported in the literature through different pathways. First, vitamin D exhibits a neurologic effect as a partial agonist on the transient receptor potential vanilloid 1 (TRPV1), which induces the desensitization of pain signals and is located in C fibers that detect burning sensations (Long et al. 2020). There is evidence that these pronociceptive receptors play a relevant role in BMS (Imamura et al. 2019). Thus, the overexpression of TRPV1 and purinergic receptors (P2) has been demonstrated along with an impairment of peripheral nerves in the pathophysiology of BMS (Adamo and Spagnuolo 2022). Therefore, Vit D Deficiency could result in an abnormal perception of sensory stimuli and increased pain sensitivity in BMS patients. Second, it has even been speculated about the relationship between low vitamin D levels, oral dryness, Candida invasion, and the sensation of burning (Gu, Baldwin, and Canning 2023). Finally, the relationship between Vit D Deficiency, mood disorders, and BMS is expanded in the subsection related to the psychosomatic network.

Conversely, the effect of BMS on Vit D Deficiency was unexpected. The authors of this study speculate about a possible indirect effect of some gastrointestinal alterations that can affect vitamin D absorption. On the other hand, besides the agreement that gastrointestinal problems can lead to a deficiency in vitamin D3, it is also acknowledged that vitamin D plays a crucial role in the absorption of magnesium and calcium (Tripathy and Majhi 2020). Its deficiency could reduce calcium absorption and even progress to hypocalcemia and hyperparathyroidism (Guarnotta, Di Gaudio, and Giordano 2022). Vitamin D deficiency could also be associated with some conditions such as diabetes mellitus, neurologic disorders, and endocrinopathies (Guarnotta, Di Gaudio, and Giordano 2022), comorbidities commonly reported along with BMS. In addition, some aspects that were related to BMS in the univariate analysis, but not in the network analysis, such as lipid metabolism alteration or the consumption of beta‐blockers, could be linked to Vit D deficiency and cortisol alterations (Guarnotta, Di Gaudio, and Giordano 2022).

4.5. Variation in Cortisol Levels

Closely related to Vit D Deficiency, the fourth associative pathway provides evidence that BMS increases the variation in cortisol levels. However, the variation in cortisol levels does not seem to directly affect BMS (Table 3). Cortisol Variation can be a connection mechanism between BMS and musculoskeletal alterations, gastrointestinal alterations, Non‐restorative Sleep, Fatigue and Cognitive problems, and Vit D deficiency, since variation in cortisol levels is shown as a common mediator between these aspects and BMS (Figure 1). The Cortisol Variation is based on a decrease in salivary cortisol levels in BMS patients that was not observed in the control group.

This increase in the variation of cortisol levels clarifies the relationship of BMS with some endocrine and metabolic changes that can lead to alterations in salivary composition and the HPA axis (Castillo‐Felipe et al. 2022; Lopez‐Jornet et al. 2020). Thus, BMS increases the variation in cortisol levels and, contributes to the potential dysregulation of the HPA axis. Indeed, dysregulation of the HPA axis could induce both elevated levels and attenuated diurnal patterns of cortisol in individuals with chronic pain (Knezevic et al. 2023), similar to those found in this study. On the other hand, an indirect influence of cortisol levels on BMS is also feasible. Even though cortisol is commonly increased in BMS patients (Fernandez‐Agra et al. 2023), the nature of this relationship is unknown. HPA axis dysregulation, often linked to chronic stress, may heighten pain sensitivity and alter metabolism, behavior, mood, and cognition (Knezevic et al. 2023), suggesting a plausible role in BMS pathogenesis.

In a broader sense, the role of cortisol as a connection mechanism with multiple physiological processes is logical (Aitken‐Saavedra et al. 2021; Fernandez‐Agra et al. 2023). In this study, one of the greatest associations is established between the variation of cortisol levels and Vit D deficiency. It is known that both glucocorticoids and vitamin D influence the immune system and show reciprocal interactions. On the one hand, vit D deficiency is associated with adrenal insufficiency, neurologic disorders, diabetes mellitus, and cardiovascular diseases, among other endocrinopathies. On the other hand, glucocorticoids indirectly reduce vitamin D levels, regulating the expression of vitamin D receptors in many tissues (Guarnotta, Di Gaudio, and Giordano 2022). The remaining relationships involving Cortisol Variation align with the mutual regulation between the activity of the HPA axis and sleep architecture. Thus, cortisol affects sleep, while changes in sleep modify the release of cortisol. Since cortisol influences sleep onset and maintenance, it also affects mood regulation during sleep. Similarly, cognitive functions depend on cortisol, with both excessively high and low levels being detrimental (Henry, Thomas, and Ross 2021). Finally, the findings of this research are also consistent with the potential effect of the HPA axis on the relationship between sleep and specific gastrointestinal alterations, such as intestinal dysbiosis (Matenchuk, Mandhane, and Kozyrskyj 2020).

4.6. Musculoskeletal Alterations

The last direct associative pathway shows a bidirectional relationship in which Musculoskeletal Alterations decrease the likelihood of BMS and vice versa. This association is predictable as Musculoskeletal Alterations generate physical and emotional stress. However, the effect is contrary to what was expected. This could be due to difficulties in the perception of musculoskeletal discomfort of the participants. These alterations were recorded also through the SSS2, which reveals similar muscle pain in both groups, as well as a predominance of muscle weakness, fatigue, and exhaustion in BMS patients. These musculoskeletal alterations are related to the variation of cortisol levels, a reasonable link since cortisol regulates the production of muscle proteins; so its alterations can generate musculoskeletal disorders (Peeters et al. 2008). Likewise, myopathies can be a consequence of low levels of vitamin D (Guarnotta, Di Gaudio, and Giordano 2022; Tripathy and Majhi 2020).

4.7. Psychological and Psychosomatic Suffering

Following the somatic network associative pathways, Psychological Suffering is indirectly linked with BMS through three aspects: Vit D Deficiency, Central Sensitization Symptoms, Non‐restorative sleep, Fatigue, and Cognitive problems.

In line with previous studies (de Pedro et al. 2020a; Kim, Kim, and Kho 2014; Schiavone et al. 2012a, 2012b; Yoo et al. 2018), Psychosomatic network shows that patients with BMS presented higher levels of Somatization, Obsession‐Compulsion, Interpersonal Sensitivity, Depression, Anxiety, Hostility, and Psychoticism than controls. However, no direct association was found between BMS and SCL‐90 psychological scales except for weak associations with Interpersonal Sensitivity and Hostility. An indirect BMS‐Hostility relationship was also observed through Cortisol Variation and Vit D Deficiency. Recent studies observed a negative correlation between Vitamin D and anger‐hostility in medical students, that is, the smaller the D vitamin levels the larger the anger‐hostility presented by the students. Additionally, deficiencies in Vitamin D have been linked to a general worsening of the emotional state or mood (Gökalp, Aydın Çil, and Yayla 2021), together with an elevated risk of experiencing depression and anxiety (Milaneschi et al. 2014); which have been closely related to mood and sleep alterations in BMS patients (Adamo et al. 2018; Castillo‐Felipe et al. 2022). The relationship between cortisol and vitamin D deficit might also be related to emotional disturbances.

A second relationship involved the indirect relation between BMS and Somatization, through Nonspecific Symptoms of Central Sensitization. Given that the somatization is associated with every other SCL‐90 scale, the central sensitization symptoms may link the deteriorated overall psychological, emotional state, and BMS. Considering both networks, central sensitization symptoms can likely reflect both psychological and somatic BMS complaints. Furthermore, cortisol, which appears strongly related to Vitamin D, is a consequence of the HPA axis activity and therefore associated with emotional and mood disturbances. As for Psychoticism, this can include unusual perceptions, odd or eccentric behavior, and nonstandard beliefs (Derogatis, Lipman, and Covi 1973). Although the association between psychoticism and BMS is described in the literature (Yoo et al. 2018), its underlying basis is rarely explored. However, BMS transcends the boundaries of mere oral discomfort, probably producing unusual perceptions (i.e., Mouth burning) and increasing eccentric behavior to reduce the mouth sensations. This might explain the larger psychoticism levels without the co‐occurrence of psychotic disorders. In this line, some studies point out that BMS might present a higher frequency of diagnosed major depressive disorder, past major depressive disorder, generalized anxiety disorder, and hypochondria, but no differences have been observed in psychotic disorders (de Souza et al. 2012).

4.8. Predictability

Despite the variety of aspects examined, the low predictability observed suggests a lack of sufficient elements to accurately classify individuals as BMS patients. Only Nonspecific Central Sensitization Symptoms, and to a lesser extent Gastrointestinal Alterations, are relatively relevant. That is, these variables show a modest effect on BMS, while they play a key role in the prediction and correct classification of cases. Therefore, their overall effect could be even greater due to interactions. It is worth noting that the variation of cortisol levels also does not show a relevant effect probably due to its nonspecific nature.

In general, this study suggests that multimorbidity in BMS patients is influenced by cortisol alterations and Vit D deficiency, with these patients often presenting systemic comorbidities such as mood disorders, hypothyroidism, chronic fatigue, sleep disturbances, etc. (Adamo and Spagnuolo 2022; de Pedro et al. 2020a; Pereira et al. 2022). However, network analysis indicates that some systemic diseases minimally impact BMS, except for gastrointestinal issues and nonspecific central sensitization symptoms, potentially linked to gut–brain interactions. Consequently, BMS may be largely independent of systemic conditions like diabetes, Parkinson's, and autoimmune diseases (Suga, Takenoshita, and Toyofuku 2020), while gastrointestinal and central sensitization conditions remain relevant comorbidities.

In summary, this study provides an associative pathways “roadmap” between some of the most relevant aspects of BMS, enhancing the understanding and management of these complex interactions. Finally, it is worth distinguishing between the associative patterns examined in this study and the pathogenesis of BMS, which is assumed to be a multifactorial process. At this point, the authors of this study speculate that the neurologic deterioration of BMS patients could be induced by the imbalance of cortisol and Vit D deficiency, in a state of central sensitization associated with multiple comorbidities such as mood alterations, gastrointestinal alterations, and other conditions that generate central sensitization, on which a stressor could trigger the symptoms of BMS from a certain threshold. Given these complex interactions, there could even be BMS endotype‐dependent phenotypes related to alterations in nociceptive processing, which could explain the different responses to treatments. These mechanisms could be related to nociplastic pain or selective nerve fibers atrophy, as has been suggested (Imamura et al. 2019).

4.9. Strengths and Weaknesses

This study includes several strengths. Notably, it is the first to explore the complex interactions involving BMS through network analysis, which enabled the identification of bidirectional effects between variables without assuming any predictors or responses a priori. It showed interactions that consider the other variables within a conservative model that facilitates interpretation and avoids overfitting. On the contrary, some weak, albeit relevant relationships could have been omitted. Other strengths include the comprehensive analysis of controversial BMS‐related factors, the use of standardized measures, and biochemical saliva records.

Some limitations of this study included its cross‐sectional design, which prevented the evaluation of causal relationships; a limited sample size, conditioned by the low prevalence of BMS, which may limit the generalizability of the findings. However, the associations were stable and reproducible, and the methodology is deemed appropriate for extracting relevant information from small to moderate sample sizes (Altenbuchinger et al. 2020). Therefore, these findings are reliable and can be used as a starting point for future research. Other limitations were the absence of neuropathic factors evaluation, and the reliance on interviews for some measures, potentially affecting accuracy. Objective measures were not feasible due to the broad range of variables analyzed. Therefore, some relevant self‐reported aspects, such as gastrointestinal alterations and Vit D deficiency, should be examined using objective measures in the future. Finally, it was not possible to examine the consumption of anxiolytics or antidepressants due to a limited sample size. The consumption of these drugs was high in BMS patients and could have decreased the observed associations, especially in psychological aspects.

5. Conclusions

This study provides a plausible framework for understanding the associative pathways between biological and psychosocial factors in BMS patients. Within this framework, gastrointestinal alterations and Vit D deficiency show a significant influence on BMS. Simultaneously, cortisol is associated with several processes of general health deterioration, as salivary cortisol levels act as a mediator in multiple associative pathways linking various psychosomatic aspects such as musculoskeletal alterations, gastrointestinal alterations, Vit D deficiency, non‐restorative sleep, fatigue, and cognitive problems. Regarding the psychological aspects, overall, BMS is identified by a widespread deterioration in emotional and psychological health that extends beyond anxiety and depression to include aspects of psychoticism, interpersonal sensitivity, and hostility. This deterioration appears to be associated with a deficiency in vitamin D. Only nonspecific central sensitization symptoms demonstrate a notable predictive capacity for BMS, suggesting the absence of relevant elements such as neuropathic factors.

Given that the main limitation of the study was the limited sample size, further studies are needed to corroborate this framework of associative pathways and clarify the nature of the relationships between BMS and prominent factors such as gastrointestinal alterations and vitamin D deficiency, as well as the role of factors not included in this framework, such as the neuropathic aspects of BMS.

Author Contributions

Oscar Gabriel Castaño‐Joaqui: conceptualization, methodology, investigation, formal analysis, visualization, writing – original draft. Laura Jiménez Ortega: conceptualization, methodology, project administration, writing – review and editing, supervision. Rocío Cerero Lapiedra: conceptualization, methodology, supervision, resources, project administration, writing – review and editing. Adelaida África Domínguez Gordillo: conceptualization, methodology, resources, project administration, supervision, writing – review and editing.

Ethics Statement

The study protocol was approved by the “Hospital San Carlos” ethics committee (reference 2.2/16).

Conflicts of Interest

The authors did not receive support from any organization for the submitted work. The authors have no relevant financial or nonfinancial interests to disclose.

Supporting information

Data S1.

Data Availability Statement

The data that support the findings of this study are available from the corresponding author upon reasonable request.