Psychopathology and the Validity of Gastrointestinal Symptom Reporting as Revealed Through Cluster Analyses of MMPI-2-RF Results

Abstract

Background

Psychological state, self-reported gut symptoms, and somatic complaints are recognized relationships that can impact health assessment and subsequent treatment.

Aim

To investigate the impact of psychological state and personality on symptom self-reporting and somatization.

Methods

Sixty-two (62) participants from the Hunter region of NSW (Australia) undertook a survey of health and lifestyle along with an MMPI-2-RF assessment of personality, psychopathology, and test-taking attitude. Participants also completed the Rome Criteria to assess functional gastrointestinal disorders (FGIDs). To assist the interpretation of MMPI-2-RF results, clustering was applied to identify similar responses and sub-cohort profiles of reporting.

Results

Cluster analysis revealed four sub-cohorts, stratified by psychopathology, gut-related symptoms, and the validity of self-reported somatic complaints. Sample clustering identified one sub-cohort defined by high rates of negative affectivity and suicidal ideation. Apart from these differences, clusters were uniform for age, sex, smoking, mental health diagnoses, as well as for gut-related conditions.

Conclusion

Results provide further evidence of the interaction of the gut–brain axis and its relationship to serious mental health conditions. It also points to the need to assess the veracity of self-reported symptomatology that may be both pathognomonic for psychopathology but might also be a consequence of gut dysbiosis. Clustering assisted these investigations by defining distinct sub-cohorts based on participant MMPI-2-RF responses.

Supplementary Information

The online version contains supplementary material available at 10.1007/s10620-024-08629-w.

Introduction

The bidirectional relationships between the gut and the brain are well established in the literature [1, 2] and an active area of contemporary research. Altered gut–brain axis signaling has been implicated in somatoform gastrointestinal disorders, but the pathophysiology of functional gastrointestinal disorders is complex and multifactorial [3, 4]. Included in this literature is evidence that supports the connection of gastrointestinal (GI) symptoms and functional gastrointestinal disorders (FGIDs) to psychological symptoms such as negative affectivity, stress, anxiety, and depression [1, 5, 6], along with personality traits, such as openness, extroversion, conscientiousness and neuroticism, and personality disorders [7]. Investigating self-reported GI factors and their link to personality and mental health presentations can provide insights into the role of somatization as a factor in the understanding of human psychology and assist to guide treatment in the clinic.

Functional gastrointestinal disorders, such as irritable bowel syndrome (IBS) and functional dyspepsia (FD), have been shown to have a clear association with psychological disorders [4, 8–11]. Links have been found between GI symptom severity and mental health disorders, such as major depressive and anxiety disorders, as well as schizophrenia and bipolar disorder [9, 12].

The impact of personality and psychopathological processes on gastrointestinal-related symptoms has been previously recognized and reported, using various editions of the Minnesota Multiphasic Personality Inventory (MMPI) [13]. This study seeks to apply the updated form of this instrument, MMPI-2-RF, in the interrogation of personality profiling in those with self-reported gastrointestinal symptoms, assisted by clustering to detect unique data structures based on similarity.

Minnesota Multiphasic Personality Inventory, second edition, revised format (MMPI-2-RF) provides a well-regarded and validated clinical tool by which it assesses personality and psychopathology [13–17]. The strength of the MMPI-2-RF is its embedded validity scales which can assist in detecting infrequent or inconsistent responding as well as the over- or under-reporting of symptoms, which can occur in self-report measures [18]. Furthermore, the MMPI-2-RF can provide objectivity to measures of somatic complaints, including gastrointestinal symptoms [19, 20]. Among gastrointestinal disorders, irritable bowel syndrome (IBS) presents a specific example to explore the validity of self-reported symptoms. IBS is noted for its diagnostic challenge whereby there is often a lack of identifiable gastrointestinal pathology in the setting of significantly described gastrointestinal complaints [21]. Helpful to the additional assessment of gastro-intestinal disorders is the Rome Criteria, which is a commonly used and established self-report measure of GI symptoms and a diagnostic tool for FGIDs [9, 10, 22, 23].

Regardless of the instrument applied, studies are only as robust as the quality of self-reporting, particularly for participants with personality disorders and/or other forms of psychological dysfunction [18]. Certain personality traits and psychopathology symptoms, such as negative affectivity, anxiety, and depression, have been associated with exaggerating and/or feigning symptoms, both physical and psychological [24]. A gap exists in the current literature regarding the participant account of symptoms when exploring the relationship between gut disturbances, personality, and psychopathology, which includes assessments of self-reported symptoms to personality types or psychopathology, and the subsequent investigation on the actual veracity of the reported GI symptoms.

The aim of this study was to investigate the relationship between gut symptoms, personality and psychopathology in a community sample, and whether the application of clusters to MMPI-2-RF results enhanced the capacity to detect exaggerated symptom reporting in this context. The impact of gut symptoms was further evaluated by the inclusion of the Rome criteria, with an in-clinic health and wellness survey also included to assess general health and identify potential confounders. The results presented herein provide guidance on assisting MMPI-2-RF interpretation, most particularly when considering the nature of the bidirectional gut–brain relationship and its implications for symptom reporting, and the accurate assessment of patient health.

Materials and Methods

Research Setting and Participation

Seventy-six (76) community participants were recruited through advertisements on community notice boards from March to July 2020 (Newcastle region, NSW). Participants completed the MMPI-2-RF, a restructured version of the MMPI-2 [14–16], as well as a health and wellness questionnaire designed by co-author Dr L. Thomas (Table S1). The MMPI-2-RF contains clinical scales that assess major categories of psychopathology and validity scales designed to assess test-taking attitudes. MMPI raw scores are transformed into standardized T-scores where the mean is 50 and the SD is 10. A T-score of 65 or greater indicates clinically significant psychopathology. Therefore, a mean MMPI score of ≥ 65 was the threshold applied to decide abnormal clinical responses. For the validity scales, a T-score of 80 or greater suggests the over-reporting of symptoms. Therefore, a mean MMPI score of ≥ 80 was the threshold applied to decide invalid responses. The MMPI-2-RF criteria included validity, higher order, and RC (somatic) response, cognitive, internalizing, and PSY-5 scales, with criteria definitions provided in Table 1.

Table 1.

MMPI-2-RF criterion scale definitions included to assess personality, psychopathology, and reporting validity for study participants

Behavior cluster	MMPI criterion	Definition
Validity scales	Fs	Infrequent somatic responses
	FBS-r	Symptom validity
Higher order and RC (somatic) scales	EID	Emotional/internalizing dysfunction
	THD	Thought dysfunction
	BXD	Behavioral/externalizing dysfunction
	RC1	Somatic complaints
Cognitive and internalizing	GIC	Gastrointestinal complaints
	SUI	Suicidal/death ideation
	STW	Stress/worry
PSY-5 scales	AGGR-r	Aggressiveness-revised
	DISC-r	Disconstraint-revised
	NEGE-r	Negative emotionality/neuroticism-revised

In addition to the MMPI-2-RF and health and wellness questionnaire, participants were required to complete the Rome Criteria [23] to assess the presence and/or impact of IBS symptoms at the time of investigation.

Cluster Analysis and Statistics

Clusters

Clustering is an unsupervised machine learning technique to detect structures within complex data via similarity. For this investigation, K-medoid (K-med) cluster analyses were conducted on the aggregated MMPI-2-RF results via the R packages factoextra and cluster (R version 4.1.1) [25–27]. K-med calculates the cluster medians, and is therefore more robust to the impact of outliers in comparison to K-means. The number of clusters (K) was optimized via the calculation of the within Sum of Square total and Gap statistic (k), both plotted against a range of clusters (Supplementary—Fig. S1). Four medoid clusters (k = 4) were applied to the final analysis. The “predict” function of the cluster was applied to assign each participant to one of four MMPI-2-RF clusters; the participant results were thereafter assigned to the appropriate sub-cohort group 1–4.

Descriptive Statistics, ANOVA and Chi-Square

Once participants were assigned to individual MMPI-2-RF cluster sub-groups, mean (± SEM) and median (range; lowest—highest MMPI-2-RF score) values were calculated. To estimate independence of the clusters, one-way ANOVA was conducted for each of the 12 MMPI-2-RF criteria, as well as age and body mass index (BMI). ANOVA results at p < 0.004 (post-Bonferroni adjustment) were considered statistically significant. Variance ratios (median with adjusted degrees of freedom) were homogeneous for age and BMI, with all MMPI-2-RF criteria homogeneous for variance, except for Fs and SUI. Statistical significance between individual sub-groups was calculated by the Bonferroni method of ANOVA post hoc analysis; only results at p < 0.001 were reported.

From health-wellness questionnaires, statistical significance was calculated via Fisher’s Exact Chi-square (χ²), with the Fisher’s exact p value reported where expected counts were less than five. The survey questions sought participant responses (yes or no) on whether they followed a special diet, experienced sleep problems and/or food intolerances, and if they had been previously diagnosed with a mental health condition. The ratio of females to males in the participant cohort was also calculated via χ², as well as responses to the Rome Criteria on gut health.

Effect sizes were calculated by Eta squared for one-way (fixed effects) ANOVA and Cramer’s V (V) statistic for Fisher’s Exact Chi-square (χ²). Effect size results allowed the post hoc calculation of statistical power for ANOVA and Chi-square analyses via G-Power [28].

ANOVA, Fisher’s Exact Chi-square (χ²), and descriptive statistical analyses were conducted via SPSS (version 27, IBM Australia).

Results

The community sample recruited for this study comprised 53 females and 23 males (n = 76) with ages ranging from 21 to 75 years, and a Body Mass Index (BMI) range from 18.1 to 51.8 (Table 2). Due to either study withdrawal or non-completion of surveys/questionnaires post-initial participation consent, the study cohort investigated via the following methods comprised 62 participants.

Table 2.

Descriptive statistics and summary of the participant cohort investigated for personality and psychopathology via the MMPI-RF-2, in the context of gut symptom profile as assessed by the Rome Criteria and self-reporting

Participant cohort summary	Mean ± SD (Range)
	Total (n = 76)	Female (n = 53)	Male (n = 23)
Age (Years)	45.35 ± 15.88	43.28 ± 14.53	50.22 ± 18.10
	(21–75)	(21–72)	(21–75)
BMI	26.18 ± 6.10	25.65 ± 6.25	27.54 ± 5.63
	(18.1–51.8) *	(18.1–51.8)	(23.8–45.2)

Gut symptoms	Proportion reporting Yes (n = 76, 77)	p value**
Rome criteria	0.632 (48/76)	0.029
IBS	0.169 (13/77)	< 0.001
Other gut disorder	0.039 (3/77)	< 0.001

^*BMI not recorded for 13 participants (8 female, 5 male)

^**Binomial test (proportions tested against 0.50)

After K-med calibration and optimization, the 62 participants were assigned to one of four cluster sub-groups based on their MMPI-2-RF responses (Fig. 1). The majority of responses fell into clusters 1 (n = 17) or 2 (n = 27), with some overlap between these sub-groups. Cluster 4 (n = 10) was clearly separated from clusters 1–2, with cluster 3 (n = 8), the most clearly differentiated from the other three sub-groups (General associations between MMPI-2-RF criteria prior to clustering are summarized by PCA in Fig. S2).

Fig. 1.

Cluster analysis (K-medoid) results derived from MMPI-2-RF responses representing 62 participants who attended the Med-Psych Clinic (Newcastle, NSW) and consented to research participation. The MMPI-2-RF profiles for each cluster are summarized in Table 3. Cluster 1 (Red); Cluster 2 (Green); Cluster 3 (Blue); and Cluster 4 (Purple)

Inspection of means and medians for each cluster identified individual MMPI-2-RF criterion that contributed to the separation of individual clusters, with mean and median MMPI-2-RF clinical scale scores ≥ 65 noted (Table 3). ANOVA analysis revealed all MMPI-2-RF criteria as significant (p < 0.001; df 3, 58–61) across the four clusters, whereas age and BMI were not significantly different (Table 3). Significant differences between individual clusters were detected by ANOVA post hoc statistics. The effect size results obtained via ANOVA for the MMPI-2-RF criteria ranged from 0.275 (AGGR-r) to 0.839 (SUI), allowing the post hoc calculation of statistical power to be estimated at between 0.102 and 0.996 (sample total of 62 participants; Bonferroni adjustment, α = 0.004). MMPI-2-RF criteria that achieved statistical power above 0.80 were FBS-r, EID, SUI, and NEGE-r, while other criteria had statistical power > 0.70 (GIC, BXD), with Fs and RC1 > 0.60. Effect sizes for Age and BMI were 0.123 and 0.144, respectively.

Table 3.

MMPI-2-RF criteria mean and medians of patient sub-populations (clusters 1 to 4) determined by K-medoids (Fig. 1)

MMPI-2-RF criteria (age & BMI)		Cluster 1	Cluster 2	Cluster 3	Cluster 4	p-valuea	Significance—individual clustersc
Age	N	17	27	8	10	0.053	NS
	Median (Range) b	41 (22–64)	50 (24–72)	29 (21–55)	51.50 (28–73)
	Mean (± SEM)	41.6 (3.0)	47.6 (3.1)	33.3 (4.3)	50.1 (5.1)
BMI (Body Mass Index)	N	15	25	5	9	0.049	NS
	Median (Range)	28 (22.4–45.2)	23.7 (18.1–36.6)	21.8 (20.4–28.3)	24.9 (23.8–39.0)
	Mean (± SEM)	28.9 (1.8)	24.3 (0.8)	23.8 (1.7)	26.3 (1.6)
Fs	N	17	27	8	10	< 0.001	3–1,2,4
	Median (Range)	50 (42–66)	42 (42–58)	70 (50–120)	50 (42–74)
	Mean (± SEM)	51.9 (1.6)	43.8 (0.8)	77.0 (8.8)	53.2 (3.0)
FBS-r	N	17	27	8	10	< 0.001	4–1,2,3
	Median (Range)	67 (51–77)	51 (35–64)	83 (67–96)	48 (35–64)
	Mean (± SEM)	65.8 (1.7)	50.7 (1.5)	81.4 (3.4)	48.7 (3.1)
EID	N	17	27	8	10	< 0.001	2–1,3,4
	Median (Range)	46 (38–62)	43 (30–61)	82 (71–86)	56.5 (45–71)
	Mean (± SEM)	48.8 (1.4)	42.7 (1.4)	80.1 (1.7)	56.5 (2.5)
THD	N	17	27	8	10	< 0.001	3–1,2
	Median (Range)	48 (39–60)	48 (39–67)	72 (48–91)	55 (48–74)
	Mean (± SEM)	46.9 (1.9)	46.9 (1.5)	68.6 (5.3)	55.7 (2.3)
BXD	N	17	27	8	10	< 0.001	1–3,4
	Median (Range)	46 (32–55)	40 (32–50)	56 (40–81)	56 (46–68)		2–3,4
	Mean (± SEM)	44.8 (1.5)	40.1 (1.0)	56.9 (4.3)	56.9 (1.8)

MMPI-2-RF criteria		Cluster 1	Cluster 2	Cluster 3	Cluster 4	p-valuea	Individual clustersc
RC1	N	17	27	8	10	< 0.001	1–2
	Median (Range)b	63 (47–77)	51 (36–63)	68.5 (56–100)	52.5 (36–61)		3–2,4
	Mean (± SEM)	62.3 (1.9)	48.9 (1.5)	72.8 (5.0)	50.9 (2.5)
GIC	N	17	27	8	10	< 0.001	1–2,4
	Median (Range)	72 (46–88)	46 (46–72)	72 (46–96)	46 (46–80)		3–2,4
	Mean (± SEM)	67.4 (3.0)	47.6 (1.2)	75.8 (5.6)	49.4 (3.4)
SUI	N	17	27	8	10	< 0.001	3–1,2,4
	Median (Range)	45 (45–66)	45 (45–66)	89.5 (79–100)	45 (45–45)
	Mean (± SEM)	47.5 (1.7)	46.6 (1.1)	89.5 (4.0)	45.0 (0.0)
STW	N	17	27	8	10	< 0.001	2–3
	Median (Range)	52 (36–73)	47 (36–65)	65 (47–81)	57 (52–65)
	Mean (± SEM)	51.7 (2.1)	45.7 (1.4)	65.8 (3.7)	57.4 (1.8)
AGGR-r	N	17	27	8	10	< 0.001	NS
	Median (Range)	45 (37–56)	45 (32–60)	37 (35–65)	56 (43–69)
	Mean (± SEM)	44.1 (1.3)	45.4 (1.3)	43.0 (3.8)	56.6 (2.8)
DISC-r	N	17	27	8	10	< 0.001	1–4
	Median (Range)	49 (35–56)	41 (35–54)	52.5 (38–72)	56 (47–79)		2–3,4
	Mean (± SEM)	46.9 (1.4)	41.9 (1.0)	54.5 (3.9)	59.2 (3.4)
NEGE-r	N	17	27	8	10	< 0.001	1–3
	Median (Range)	47 (40–66)	45 (32–51)	73 (56–80)	56 (47–73)		2–3,4
	Mean (± SEM)	48.3 (1.6)	43.2 (1.2)	70.5 (3.2)	57.5 (2.3)

Statistical significance determined by one-way ANOVA, including post hoc comparisons and effect sizes. Cluster MMPI clinical scores ≥ 65 T and validity scores ≥ 80 T (see Table 4 for the percentage of the cluster sample elevated ≥ 65)

^aANOVA p value (Cluster 1–4)

^bRange (Lowest to Highest for each cluster)

^cPost hoc comparison method—Bonferroni

NS not significant

The result that emphasized cluster differentiation were the following clinical scales elevated for cluster 3—Suicidal/Death Ideation (SUI), Emotional-Internalizing Dysfunction (EID), Thought Dysfunction (THD), Somatic Complaints (RC1), Stress—Worry (STW), Negative Emotionality—Neuroticism (revised) (NEGE-r), and Gastro-Intestinal Complaints (GIC) (Table 3). Cluster 3 mean and medians greater than 80 were also noted for symptom validity (FBS-r). Only clusters 1 and 3 reported a mean/median > 65 for gastro-intestinal complaints (GIC) with commensurate statistical significance (p < 0.001).

To further evaluate the differences between cluster profiles, the percentage of participants who scored at or above 65 (≥ 65) were calculated for each cluster and MMPI-2-RF criterion (Table 4). As noted above (Table 3), only clusters 1 and 3 recorded means/medians ≥ 65, and this observation was reflected by these percentages for the same criteria. From the 12 criteria measured for this study, cluster 1 had five with zero cases of ≥ 65, and cluster 2 had 8 not recording an elevation above this threshold. Furthermore, of the criteria that recorded elevated cases for cluster 2, all were < 10% of the cohort. Cluster 4 recorded 4 criteria without an elevated case, noting STW, AGGR-r, DISC-r, and NEGE-r as the highest of the elevated criteria at between 20 and 40% of this participant cohort. Again, cluster 3 was remarkable in relation to the MMPI-2-RF results. All criteria recorded elevated cases, with 100% (8/8) of cluster participants at or above 80 for validity scale FBS-r, at or above 65 observed for clinical scales EID and SUI, and other exceptional criteria ranging between 63 and 88%, corresponding with the observations in Table 3. The lowest percentage was observed for AGGR-r (1/8).

Table 4.

The percentage of cluster 1–4 sub-cohorts (community participants) who recorded a score of ≥ 65 T on each of the clinical MMPI-2-RF questionnaire criteria

MMPI-2-RF criteria*	Percentage of sample ≥ MMPI-2-RF 65 T
	Cluster 1 (n = 17)	Cluster 2 (n = 27)	Cluster 3 (n = 8)	Cluster 4 (n = 10)
EID	NA	NA	100% (8/8)	10% (1/10)
THD	NA	3.7% (1/27)	63% (5/8)	10% (1/10)
BXD	NA	NA	25% (2/8)	10% (1/10)
RC1	29% (5/17)	NA	75% (6/8)	NA
GIC	53% (9/17)	3.7% (1/27)	88% (7/8)	NA
SUI	12% (2/17)	7.4% (2/27)	100% (8/8)	NA
STW	5.9% (1/17)	3.7% (1/27)	75% (6/8)	30% (3/10)
AGGR-r	NA	NA	12.5% (1/8)	40% (4/10)
DISC-r	NA	NA	25% (2/8)	30% (3/10)
NEGE-r	5.9% (1/17)	NA	75% (6/8)	20% (2/10)

^*MMPI-2-RF definitions (Table 1)

NA no clinical scores at or above 65 T (≥ 65)

For the understanding of gastro-intestinal influences on participant responses, in addition to the GIC criterion from the MMPI-2-RF, parallel Rome Criteria responses were collected to evaluate the impact of IBS on symptom reporting. All eight participants in cluster 3 reported affirmative (Yes) responses, confirming IBS symptoms at the time of investigation, whereas cluster 1 reported 9 from 16 “Yes” responses under the Rome criteria. No statistical differences between clusters 1 and 4 were detected by Chi-square (p = 0.14; df = 3, 57; Effect size = 0.301) when comparing Rome Criteria responses (Table 5). To consider the influence of other health features on symptom reporting, Chi-square analyses on clusters 1–4 were also conducted to assess sleep problems, food intolerances, adherence to special diets, and reports of previously diagnosed mental health conditions (Table 5). No significant differences were detected for these features (all p > 0.25; effect sizes—low (0.055) to moderate (0.250)), with the proportion of females to males also found not to be significant across MMPI-2-RF clusters (p = 0.32; Effect size = 0.246).

Table 5.

The number of affirmative responses to the Rome Criteria and questions from an in-clinic health and wellness questionnaire returned by study participants who completed the MMPI-2-RF

Health–Wellness Questions (Yes or No response)	Affirmative responses for each cluster (total responses)				p value Fisher’s Exact test
	Cluster 1 (16)	Cluster 2 (27)	Cluster 3 (8)	Cluster 4 (10)
Rome criteria*	9	18	8	8	0.14
Female	11	20	4	9	0.32
Special diet	9	15	2	7	0.31
Sleep problems	5	6	3	3	0.78
Food intolerance	4	6	1	3	0.89
Diagnosed mental condition	8	12	4	5	0.98

Clusters represent MMPI sub-groups determined post K-medoid cluster analysis (See Table 3)

^*Rome Criteria—a separate inventory for self-reported gut symptoms

As noted, the interpretation of self-reporting across the survey instruments employed requires closer consideration for the cluster 3 sub-cohort, as indicated by the FBS-r validity scale responses (Table 3), before firm conclusions can be drawn.

Discussion

This study identified associations between participants who over-report gastro-intestinal complaints. The implication of over-reporting threatens the validity of research that uses self-report measures for somatic symptoms. To investigate these subjective gastrointestinal complaints, cluster (K-medoid) analysis was applied to MMPI-2-RF results provided by 62 participants. The optimized cluster analysis (Fig. 1) revealed 4 distinct sub-cohorts ranging from 8 to 27 participants per cluster. Once separated into clusters, mean and median MMPI-2-RF scores were calculated for each cluster (Table 3), with categorization for Rome criteria along with responses to a health—wellness questionnaire also defined by individual clusters (Table 5).

Clusters 2 and 4 mean/median MMPI-2-RF scores were not elevated above 65, with cluster 1 having > 65 for GIC only. Cluster 3 was characterized by clinically significant MMPI-2-RF scores for emotional and thought dysfunction, somatic and gastrointestinal complaints, and suicidal ideation. However, caution is required for cluster 3 MMPI-2-RF results as scores were also elevated for the validity scale FBS-r, which indicate infrequent reporting and somatic and cognitive complaint over-reporting. These issues in reporting somatic and GI complaints question the validity of self-reported GI symptoms from participants with certain personality profiles and/or psychopathology, and level of somatization versus physically explained GI symptoms. The MMPI-2-RF results for cluster 3 participants support research by Aparcero et al. [29] that showed participants with personality and psycho-pathological challenges are more likely to feign or exaggerate their symptoms compared to healthy participants.

One possible explanation for our results is that personality profiles and/or psychopathological disorders, characterized by low tolerance for distress and excessive rumination of problems, will distort symptom reporting [30–32]. Somatization, meaning psychological distress manifested as physical symptoms, have been found to exist in research measuring linkages between GI symptoms and MMPI assessed personality and psychopathological symptoms [13]. Multiple studies have found somatic disorders to exist in Rome Criteria-assessed FGIDs [33–35] but have not questioned the validity of the self-reported GI symptom scales to determine the level of somatization versus physically based symptoms.

The MMPI-2-RF’s validity scales offer an objective measurement of self-reported GI symptoms and somatization [36], which was extended by this investigation through understanding MMPI-2-RF response structures via cluster analysis and whether response validity requires deeper consideration. A meta-analysis [29] found that MMPI-2-RF validity scales effectively separated accurate participant responses from participants who feigned or exaggerated psychological or medical symptoms. Specifically, the embedded Fs validity scale detects infrequent somatic responses in contrast to the FBS-r validity scale that detects non-credible symptoms [29].

In relation to GI symptoms, emotional dysfunction, such as stress, anxiety, and depression, have been observed to exacerbate GI symptoms [1, 2]. Because of psychological stress’s influence on FGID development and GI symptoms, it is essential to detect somatization early to prevent or reduce the exacerbation of GI symptoms [1, 37] and to guide treatment decision-making. Future research should assess self-reported GI symptoms in conjunction with measures of somatization [29] and consider psychological interventions as early treatment for FGIDs [37].

Our study found that the report of GI symptoms correlates with certain (high order) dysfunction types and with specific symptoms of psychopathology, but that distress may need to be interpreted with caution because of over-reporting. The cluster results emphasized the need to stratify MMPI-2-RF results that had the greatest elevation of scores pertaining to both IBS and psychopathology and as suggested by clinical history, were also more likely to have an altered gut homeostasis because of other factors, such as allergy, autoimmune disease, or the prescribing of psychotropic medication (results not shown).

While physiological GI tests, such as microbiome testing, can directly explore questions of causation and thereafter propose appropriate treatments for GI symptoms and FGIDs [1, 38], GI and somatic symptoms and disorders are often measured via self-report surveys because they are a more accessible and affordable option [23, 39]. Therefore, it is essential that self-reporting is optimized. The results of this investigation emphasized issues in self-reporting as determined via MMPI-2-RF validity scales revealed by interrogating the cohort data structures by K-med cluster analyses. In short, the physical complaints reported by some participants may have no identifiable physiological foundation and as such need to be considered when assessing participants. This may ultimately include the role of somatization in increasing the rate of gastrointestinal complaints in those with psychopathology or personality disorders.

The primary limitation of this investigation was the small number of total research participants that resulted in smaller sample sizes for the separate clusters that were the focus of deeper examination. This was a particular problem for cluster 3 with only 8 participants but representing several MMPI-2-RF features of interest to our aims. Small sample size introduces the risk of inadequate statistical power (1 − β), which for the ANOVA results (Table 3) were often compensated for by large effect sizes. However, despite large effect sizes, some ANOVA results did not achieve statistical power of greater than 0.80 (range: 0.102 to 0.996), suggesting caution for the interpretation for some results. For noting within this limitation, Standard Error of the Mean (SEM) values were less than 10% of the mean across all clusters, except Fs in Cluster 3. Cluster sample size was also a challenge for the statistical investigation of the Rome criteria and the role of other self-reported health factors (Table 5). To address this limitation Fisher’s exact probability was calculated, rather than the p value estimated via standard Chi-Square.

As well as sample size, the cohort investigated represented a community convenience sample that may suggest self-selection bias. To address these limitations the study relied upon the application of validated measurements of personality/psychopathology (MMPI-2-RF) and gut disorders (Rome Criteria) to a new Australian participant cohort, whose assessment also included a clinical questionnaire to detect biases. To augment the rigor provided by these validated instruments, an unsupervised cluster algorithm was applied to define distinct MMPI-2-RF response sub-cohorts and link these to other measures of GI and general health, with implications for the deeper understanding of factors that cause gut dysbiosis, and to ultimately assist the treatment of mental health disorders in the context of self-reporting veracity independent of location.

The results of this investigation provide a quantitative method to reveal patterns contained within MMPI-2-RF response profiles that will assist in the evaluation of reporting veracity and therefore improved health assessment. In this context, the embedded validity scale FBS-r was valuable and linked to suicidal ideation (SUI), as well as other markers of psychological distress. The relationship between suicidal distress and IBS has been previously noted [40, 41]. This distress is perceived as significant by the responder and importantly, not only impacts the way in which symptoms are related, but also the behavioral response to those symptoms.

Considering these results, we speculate that gut dysbiosis can impact personality, which leads to low tolerance for distress and heightened somatic focus on gut symptoms. This was reflected by cluster 3, which combined elevated responses for a number of psychological—personality dysfunctions, gut symptoms (GIC), and also elevated responses to internal validity scales within the MMPI-2-RF. Applying the cluster methods developed herein, subsequent studies will request fecal samples from participants to screen for gut dysbiosis, to explore whether the results obtained via survey questionnaires and psychological instruments are reflected physiologically via microbiome aberration.

Supplementary Information

Below is the link to the electronic supplementary material.

Funding

Open Access funding enabled and organized by CAUL and its Member Institutions. No funding was received for this project.

Data availability

Available on request, directed to the corresponding author.

Declarations

Conflict of interest

Linda Thomas is the Director of Clinical Psychology at Med-Psych. Paris Lang was a casual employee of Med-Psych during the inception and later phases of this project.

Ethical approval

Human ethics approval was granted by the ANU Human Research Ethics Committee (ANU-HREC: protocol 2018/625). Only participants who read and agreed with the research participant information, and were willing to sign the attached consent form, were recruited for this study.