A Framework for Clinical Trials of Neurobiological Interventions That Target the Gut-Brain Axis in Inflammatory Bowel Disease

Abstract

A growing body of evidence from preclinical, translational, and clinical studies supports a bidirectional relationship within the gut-brain axis that contributes to neurobiological symptoms including anxiety, depression, fatigue, stress, and sleep disturbance. These symptoms have a significant impact on health-related quality of life and functional ability in individuals with inflammatory bowel disease. Clinical studies that generate high-quality evidence on pharmacological and nonpharmacological (eg, psychosocial, behavioral) interventions are needed to ultimately improve access to safe and effective therapies that have a meaningful impact on patients and to guide medical and regulatory decisions. This review outlines a framework for designing and conducting randomized controlled trials for interventions that target neurobiological symptoms in patients with inflammatory bowel disease based on the most recent guidance published within the past 5 years from policy makers, clinicians specialized in inflammatory bowel disease, patient-reported outcomes methodologists, health economists, patient advocates, industry representatives, ethicists, and clinical trial experts.

randomized control trial, clinical trials, neurobiological symptoms, psychological symptoms, anxiety, depression, fatigue, stress, sleep, patient-reported outcomes, endpoints

Introduction

The pathogenesis of inflammatory bowel disease (IBD) involves genetic susceptibility, environmental factors, gut microbiome dysbiosis, and immune dysregulation, although the exact mechanism remains to be elicited.¹ A growing body of evidence from preclinical, translational, and clinical studies in patients with IBD suggest that dynamic bidirectional interactions within the gut-brain axis contribute to neurobiological symptoms such as stress,² depression,^{3, 4} anxiety,³ fatigue,^{5, 6} and sleep disturbance.⁷ These neurobiological symptoms have significant negative impacts on an individual’s functional ability, emotional well-being, and quality of life and contribute to the overall disease burden in IBD with an increased risk for disease relapse and health care utilization.^8-11

Brain-to-gut signaling through neuroendocrine transmission between the enteric, autonomic, and central nervous system in IBD is supported by preclinical and clinical observations detailing the effects of psychological distress on proinflammatory cytokines (eg, tumor necrosis factor-alpha, interleukins),¹² intestinal permeability,^{13, 14} and the intestinal microbiota composition in patients with IBD.^15-17 Gut-to-brain signaling in IBD is supported by evidence from preclinical and clinical observations that intestinal inflammation, intestinal microbiota composition, and structural changes to the central nervous system in patients with IBD affect neurocognitive function and psychological well-being, including mood disorders (eg, anxiety, depression),^{3, 18} alteration of circadian rhythm (eg, sleep disturbances),^{6, 19} fatigue,⁵ and pain perception.³

Pragmatic clinical research that investigates the extent to which pharmacological and nonpharmacological interventions impact the overall disease burden of IBD and neurobiological symptoms from brain-gut interactions is urgently needed.^{11, 20, 21} Clinical studies should be designed in a way that isolates the effect(s) of the intervention from other influences, such as “spontaneous change in the course of the disease, placebo effect, or biased observation.” ²² Although meaningful changes in quality of life and neurobiological symptoms can be patient-reported (eg, have heterogeneous expression and interpretation) and can be influenced by the natural variability of the disease (particularly with the relapsing and remitting nature of IBD), the goal of clinical trial design is to construct efficient and effective studies within these research constraints to support the development of treatments that are better tailored to patients’ needs.²³

Patient-reported outcomes (PRO) are now at the forefront of determining treatment efficacy in clinical studies because it is important to include the perspectives of those living with a condition in the determination of meaningful endpoints.^{23, 24} The U.S. Food & Drug Administration (FDA) mandates that “symptoms or other unobservable concepts known only to the patient can only be measured by PRO measures,” which are reported “directly from the patient (i.e., study subject) about the status of a patient’s health condition without amendment or interpretation of the patient’s response by a clinician or anyone else.” ²⁵

This architectural article aims to provide a framework for approaching clinical trials of pharmacological and biopsychosocial therapies that target modifiable neurobiological factors in the gut-brain axis in IBD to potentially aid investigators, pharmaceutical companies, and granting agencies to “help ensure that patients’ voices are central to informing shared decision making, labeling claims, clinical guidelines, and health policy, making patient-centered care a reality.” ²⁴ The content of this article reflects the most updated guidance from the International Council for Harmonisation for contemporary clinical study design, which is endorsed by the FDA²³ and the European Medicines Agency²⁶; the international consensus-based Guidelines for Inclusion of Patient-Reported Outcomes in Clinical Trial Protocols²⁴; and the most recent guidance for clinical trial designs and PRO instruments published within the past 5 years.^{22, 23, 25, 27-32}

STRUCTURAL DESIGN, INTERVENTIONS, AND CONTROL GROUPS

Study Design

Single-arm trials, placebo-controlled trials, crossover trials, active comparator trials, and comparisons with no treatment are examples of clinical trial designs used to test the efficacy of an intervention depending on the research objectives. ²³ Interventions in clinical trials that target neurobiological symptoms in IBD can be broadly classified into 2 categories: pharmacological and nonpharmacological. Fig. 1 illustrates the factors to consider in the design of a clinical trial for interventions of neurobiological symptoms in individuals with IBD depending on the research objective, the specific endpoints of interest, and the target population.

Figure 1.

Framework for clinical trials for interventions targeting neurobiological symptoms in individuals with IBD. Outlines 8 key elements to consider in the design of clinical trials for interventions that target neurobiological symptoms in individuals with IBD. The investigators’ research objectives and hypotheses impact the methodological approach, target population, instruments for data collection, endpoint determination, and data analysis.

Interventions

The majority of pharmacological interventions in clinical trials that target neurobiological symptoms in IBD have involved antidepressants³³ (eg, duloxetine, fluoxetine, buproprion³⁴), although some trials have investigated the impact of biologic therapy on neurobiological symptoms as secondary outcomes.³⁵ In contrast, many forms of nonpharmacological therapies have been studied including behavioral and cognitive therapies,^{34, 36-39} dietary supplementation (eg, ferric maltol, omega-3, Agaricus blazei Murrill-based mushroom extract),⁴⁰ and complementary and alternative medicine (eg, acupuncture, yoga, exercise regimens).^40-42 The most common form of behavioral/cognitive intervention that has been studied to date is cognitive-behavioral therapy (CBT), in which maladaptive thinking such as negative illness beliefs (eg, “My illness has complete control over my life”) and dysfunctional attitudes (eg, “I have to be perfect to be happy”) are reduced through interventions such as behavioral activation, response prevention, imagination, and reframing.⁴³ Other forms of psychotherapeutic interventions studied to date include problem-solving, stress management, meditation, guided imagery, hypnotherapy, and yoga with mixed results in patients with symptomatically quiescent IBD.^{36, 40, 42}

More recently, 2 newer forms of psychotherapeutic interventions with positive effects on stress and fatigue in individuals with IBD were published: acceptance and commitment therapy, in which an individual is encouraged to accept the existence of negative thoughts or feelings (ie, to foster “psychological flexibility” rather than to eliminate these thoughts or feelings as with CBT),⁴⁴ and stepped-care intervention with behavioral therapy and thereafter pharmacologic therapy (ie, using bupropion, an antidepressant) for the subset of patients with persistent neurological symptoms.³⁴ These results have not yet been replicated but are promising avenues for further research.

Control Groups and Comparators

Standard-of-care and waitlist control groups

The Council for International Organizations of Medical Sciences recognizes that “the use of placebo is usually uncontroversial in the absence of an established effective intervention” and that researchers do not “delay or withhold an established effective intervention from participants.” ⁴⁵ In clinical trials that target neurobiological symptoms in patients with IBD (eg, fatigue, anxiety, depression, and stress), the most common comparison groups for psychotherapeutic and behavioral interventions have been (1) a standard-of-care control group, also called best medical management, usual care, or treatment as usual,^{8, 36, 37, 40, 42, 44, 46} and (2) a waitlist control group, in which participants receive the standard of care with the option of receiving the intervention at the end of the study period (Fig. 1).^{8, 36, 39, 40, 42, 43}

Goren et al³⁹ recently published in 2021 an example of a randomized parallel-group physician-blinded controlled trial with a waitlist control group: 139 patients with mild-to-moderate Crohn disease activity were randomized to participate in a 3-month program of cognitive-behavioral and mindfulness-based stress reduction using daily exercise taught weekly via one-on-one videoconferences with social workers (n = 70 intervention arm) or to a waitlist control group (n = 69). Waitlisted participants received treatment as usual until the end of the study period (completion of the last follow-up outcomes assessments at 3 months) and then had the opportunity to participate in the stress-reduction program.³⁹

Although some researchers may regard waitlist control groups as being more equitable for study participants than standard-of-care control groups in psychological and behavioral intervention studies, drawbacks that should be considered include program administration costs/time (higher/more in waitlist control groups than in standard-of-care control groups) and the potential risk of overestimating the active treatment effects. Waitlisted patients may be less inclined to engage in self-help behavior they would otherwise engage in if they did not have the expectation of the intervention at the end of the study, and the anticipation of the intervention may confound the outcomes (report less improvement than standard-of-care control groups).⁴⁷

Attention, sham, and placebo control groups

Several psychotherapeutic trials in IBD have used attention control groups as a form of sham comparator where participants in the control group receive the same frequency and duration of interpersonal interaction with research personnel and are engaged in comparable control activities (eg, educational lectures on IBD) as the intervention group (eg, mind-body psychotherapy; Fig. 1).^{37, 48, 49} Attention control groups in behavioral intervention trials can in theory reduce biases from the Hawthorn effect, where participants’ responses are consciously or unconsciously influenced by attention from investigators (“showering of medical attention”).^{46, 50, 51} Other forms of placebo or sham control groups are uncommon, but 1 trial on the effects on fatigue in patients with IBD compared electroacupuncture with a sham procedure (needles were inserted in “non-acupuncture” locations on the body without an electric current).⁴¹

Placebo controls are most common in randomized controlled trials (RCTs) of pharmacotherapies to evaluate their impact on anxiety,³³ depression,³³ and fatigue in patients with IBD (Fig. 1).⁴⁰ Placebos are inert substances with a form and administration indistinguishable from the active drug.⁴⁵ Active substances that have been studied include adalimumab,³⁵ antidepressants,³³ and nutritional supplements.⁴⁰ Active comparator trials, in which an investigational intervention is compared to an established effective intervention in the presence of credible uncertainty about its superiority (referred to as “clinical equipoise”),⁴⁵ where neurobiological symptoms in patients with IBD were assessed as the primary outcome, have not been published to date, likely because of the lack of high-quality evidence-based studies already published for interventions with confirmed treatment efficacy on anxiety, depression, fatigue, stress, and sleep disturbance in the IBD population.^{8, 36, 52, 53}

PATIENT SELECTION AND SAMPLE SIZE DETERMINATION

Target Population, Inclusion and Exclusion Criteria

Defining the target population in a study is important for endpoint determination and result analysis and interpretation, with implications for the relevance, reliability, and generalizability of a study’s observations.³² To illustrate this point, the FDA offered an example where “patients with late-stage disease may have different symptoms or perspectives than patients in the early stage.”³¹ Because patient selection for a particular study depends on the research objective and resource constraints, input should be sought from patients living with the condition, statisticians, and psychometricians when designing a clinical study.^{31, 32}

Prototypical inclusion and exclusion criteria that could be considered when designing clinical trials for interventions on neurobiological symptoms in individuals with IBD are illustrated in Table 1. Factors to consider include IBD phenotype and disease activity (based on symptoms alone or with objective inflammatory burden), neurobiological symptom burden (eg, cutoffs on PROs; mild, moderate, or high symptomatic ratings), and psychiatric history (including medications and treatments).^{24, 36, 37} Depending on the research objective and the attributes associated with the endpoints of interest, the inclusion and exclusion criteria need to reflect “the characteristics of the target population, the disease or condition under study, and the intended use of study results” so that statements made about the qualitative/quantitative data are meaningful and relevant to the target population (Table 1).³²

Table 1.

Prototypical Inclusion and Exclusion Criteria to Consider for Clinical Trials That Target Neurobiological Symptoms in Individuals With IBD

Factors	Inclusion Considerations	Exclusion Considerations
Age (y)	≥18	< 18
IBD phenotype, disease activity, and medications	IBD subtypes: Crohn disease, UC; prior IBD-related surgeries (eg, colectomy, small-bowel resection); diagnostic criteria by clinical, endoscopic, and histologic confirmation	Indeterminate colitis; restriction of IBD disease activity (symptomatic, biochemical, endoscopic); changes in IBD medications in the past 3 months; concurrent corticosteroid use (potential psychological and endocrinological effects)
Psychological status and psychiatric history	Past participation in psychotherapeutic interventions; stable doses of antidepressants (dose or type) within last 3 mo; clinically significant baseline neurobiological symptoms (eg, cutoff scores on baseline psychological screening)	Concurrent participation in psychotherapeutic intervention; major psychiatric disorders per Diagnostic and Statistical Manual of Mental Disorders, 5th ed.; active alcohol, opioid, or illicit drug dependency
Medical and social history	All relevant race/ethnicity, level of education, socioeconomic status	Comorbid major medical illness that would require >3 medical visits/y, including severe cardiac or pulmonary disease, renal disease requiring dialysis, decompensated cirrhosis, and cancer; pregnant at time of enrollment
Literacy, access, and cognitive abilities	Internet access and access to computer or mobile smartphone/tablet (eg, web-based assessments)	Difficulty understanding English; numeracy skills (eg, quantitative scale assessments); physical or logistical limitations for in-person interventions/testing; cognitive impairments (dementia, intellectual disability)

Because there is a poor correlation between symptomatic disease activity (eg, the Harvey-Bradshaw Index, the Simple Clinical Colitis Activity Index) and objective assessments of inflammation (eg, biochemical and endoscopic),⁵⁴ the inclusion of biochemical and/or endoscopic evaluations (eg, fecal calprotectin, serum C-reactive protein, ileocolonoscopy) in the study design for enrollment, allocation, and outcomes assessment may offer additional insights into the impact of pharmacological and nonpharmacological interventions on the psychological comorbidities of somatization, anxiety, depression, fatigue, and stress^39,54 (Fig. 1 and Table 1).

In individuals with IBD, the IBD subtype (primarily Crohn disease or ulcerative colitis [UC]) and disease activity (active vs quiescent) are some of the most prominent factors that delineate patterns of health care utilization, medical management, and neurobiological symptom burden.^{5, 8, 53, 55} The majority of RCTs conducted to date that have investigated psychotherapeutic interventions for neurobiological symptoms have been limited to patients with symptomatic remission.^{8, 36, 40, 42} This trend does not fully reflect the heterogeneity of disease activity and renders a knowledge gap about the extent to which adjunctive interventions that target neurobiological symptoms impact patients with symptomatically active disease alongside the use of dedicated pharmacologic therapies directed at reducing the inflammatory burden.^{2, 53, 55} The additional challenge to studying patients with active disease also needs to be recognized, however, because the use of concomitant medical therapies may confound outcomes assessment.³⁷

Depending on the research objective, acceptable methods to delineate the effects of “attributes that are associated with the endpoints of interest” and approximate heterogeneity in the target population include (1) a subpopulation study, where the study sample is restricted to participants with characteristics of interest (eg, active vs quiescent disease), or (2) a subgroup analysis of a broader target population with a prespecified sample size determination to facilitate “weighing to account for the over-sampling (or under-sampling) of certain subpopulations if the sample is obtained with known sampling probabilities.” ³²

An example of a recent subpopulation study (quiescent disease activity) is a feasibility RCT published in 2019 in which the effects of telephone-delivered CBT (intervention) on fatigue (measured by the IBD-Fatigue scale⁵⁶) were compared between patients with clinical-biochemical quiescent IBD (measured by the Harvey-Bradshaw Index or the Simple Clinical Colitis Activity Index and the fecal calprotectin level within 3 months of study enrollment) and control patients who received a 4-page information sheet on fatigue in IBD (ie, without interaction with therapists).⁵⁷

An example of subgroup analysis (UC vs Crohn disease) is the INSPIRE study by Boye et al⁵⁸ that compared the impact of stress management psychotherapy on the quality of life and risk of relapse in patients with IBD. A total of 58 patients with UC and 56 with Crohn disease were randomized to participate in 6 to 9 weekly sessions of psychoeducation and CBT (intervention) or continue treatment as usual (control patients). Mixed linear model analyses were performed separately for UC and Crohn disease to delineate the intervention’s effects on the 2 IBD subtypes over the course of 18 months of follow-up. Patients with UC, as compared to Crohn disease, seemed to have more improvements in the disease course and health-related quality of life. However, this analysis was performed posthoc rather than being prespecified in the study design.⁵⁸ Where there is interest in subgroup analysis (eg, UC vs Crohn disease) within the target population (eg, IBD), subgroup analysis should be determined a priori and the sample size should be calculated to ensure adequate representation from the subgroups.³²

Sample Size Calculations

Sample size calculations allow investigators to determine the number of study participants to enroll to provide adequate clinically relevant data for statistical power²³; however, enrolling too many patients can be unnecessarily costly or time-consuming.⁵⁹ The number of patients should be sufficient to make statistical conclusions with adequate precision to detect treatment effect and to control the probabilities for false conclusion.²³ A type I error rate (α) of 0.05 for 2-sided tests (ie, a 5% probability of falsely concluding that the intervention is different from the control when no difference exists) and 80% power (calculated as 1-β, where β is the probability of committing a type II error; ie, an 80% probability of identifying a treatment effect when indeed a true treatment effect exists and a 20% chance that a significant difference is missed) are most commonly used for sample size calculations in clinical trials.²⁸ Sample size calculations should also take into account dropout and nonresponse rates to arrive at a recruitment target (depending on how missing data are accounted for).³² Past studies of psychosocial and pharmacological interventions targeting neurobiological symptoms in patients with IBD have had attrition rates of 0% to 50%, depending on the mode of intervention (eg, time commitment, travel requirements), duration of study (more participants are lost to follow-up as the study lengthens), and study setting (eg, in person, telephone, or videoconference; real-time or asynchronous; one-on-one or groups).^{21, 33, 34, 37, 60}

When a PRO is the primary outcome, the target sample size is generally based on an a priori sample size calculation for that endpoint.²⁴ The minimal clinically important difference (MCID) of the primary outcome can aid in the determination of the sample size (the number of patients needed in each treatment group to detect meaningful clinically significant differences).⁶¹ The MCID is a patient-centered concept that determines the magnitude of improvement meaningful to patients, because small differences (especially in large studies) may be statistically significant but lack clinical relevance from the perspectives of patients and clinicians.^{30, 62} The most common approaches for determining MCIDs for PRO measures are consensus, anchor, and distribution-based methods, although the details of MCID determination are beyond the scope of this article.^{61, 63} The FDA recommends anchor-based methods to establish a range of score change that represents meaningful within-patient changes (rather than group-level data).³⁰ An example of anchor-based and distribution-based methods for determining the MCIDs for health-related quality-of-life outcome measures in IBD have been outlined for the GEMINI II and GEMINI III vedolizumab registration trials.⁶⁴

SELECTION OF PRO ENDPOINTS

Clinical benefit is defined as a positive clinically meaningful change on how an individual feels, functions, or survives; these elements are measured in research studies using clinical outcome assessments reported by patients, clinicians, or observers or are assessed by performance measures.²⁵ The scope of this article focuses on PROs for anxiety, depression, sleep quality, stress, and fatigue, which are important contributing factors to health-related quality of life and disability among patients with IBD^{5, 9, 12, 15, 16, 52} and are regarded as “long-term treatment targets irrespective of other objective markers of inflammation” according to the recently updated STRIDE-II study.²⁰

The National Institutes of Health Patient-Reported Outcomes Measurement Information System (PROMIS) has developed more than 300 valid, responsive, and general (not disease-specific) PRO instruments for measuring the symptoms, function, well-being, and general health of individuals living with chronic conditions and in the general population (detailed information available at https://www.healthmeasures.net/explore-measurement-systems/promis).⁶⁵ One major advantage of using PROMIS measures is that they were developed using the modern standards for systematic and rigorous PRO development methodologies, including literature reviews, qualitative item reviews, focus groups, cognitive interviewing, and psychometric testing. All PROMIS measures are interpreted using a T-score of 50 and an SD of 10 calibrated to the mean of the U.S. general population.^{65, 66} Higher scores indicate more symptoms within a domain. PROMIS scores < 55 are considered within normal limits, scores from 55 to 60 as mild, scores from 60 to 70 as moderate, and scores >70 as severe symptoms of the domain measured.^{66, 67} The MCID is considered as having a T-score of 2.5.⁶⁸

Although PROMIS measures for depression, anxiety, fatigue, and sleep disturbances are available (Table 2) and have been studied for construct validity and responsiveness in a large sample of patients with IBD in the United States (10,634 patients with IBD: 6689 with Crohn disease and 3945 with UC/indeterminate colitis),^{69, 70} they have not been widely used in research studies among patients with IBD. Moreover, the FDA has not endorsed specific PROs for determining treatment efficacy in neurobiological domains in patients with IBD but rather encourages “early discussions with the FDA about outcome measure selection.”^{29, 31, 71} For example, in 2019, the FDA accepted the use of the 10-item PROMIS Short Form Fatigue as a determination of efficacy studies of adults with rheumatoid arthritis after it was first entered into the FDA’s Clinical Outcomes Assessment qualification program in 2016.²⁷ Examples of the most commonly used PROs for neurobiological symptoms in observational and interventional studies in patients with IBD are illustrated below and in Table 2.

Table 2.

Validated PRO Measures for Depression, Anxiety, Stress, Fatigue, and Sleep Most Commonly Used in Studies of Individuals With IBD

PRO Measurement	Description (estimated time to complete)	Interpretation	Estimated Scores in Patients With IBD	Remarks
Depression
PROMIS Depression65	8 items, each scored 1-5, total score 38.2-81.3 (transformed scores into T-scores); recall period: past wk (2 min)	Higher scores indicate more depressive symptoms. T-score = 50 and SD = 10 represent the general population; T-score > 50 is abnormal, 2.5 considered as MCID65, 68	IBD T-score = 54 (SD = 10, n = 10,634); UC = 54 (SD = 10, n = 6689); CD = 54 (SD = 10, n = 3945)69	Developed by the NIH; validated in IBD69, 77
HADS-Depression76	7 items, each scored 0-3, summed score 0-21; recall period: past 7 d (2 min)	Higher scores indicate more depressive symptoms; cut points 8, 11, and 15 for mild, moderate, and severe depression	IBD mean = 5.4 (95% CI, 4.9-5.8; n = 7589)94; CD = 6.09 (95% CI ± 0.89; n = 1299); UC = 5.5 (95% CI ± 1.08; n = 1694)53	Most commonly used in patients with IBD94
BDI-II81	21 items, each scored 0-3, summed score 0-63; recall period: past 2 wk (5-10 min)	Higher scores indicate more depressive symptoms; cut points 14, 20, and 29 for mild, moderate, and severe depression	IBD mean = 9.1 (95% CI, 8.6-9.5; n = 398)94	Second most commonly used scale in patients with IBD94; emphasis on somatic symptoms (eg, loss of appetite, fatigue, loss of energy) may overestimate depression in medically ill patients77
PHQ-978	9 items, each scored 0-3, summed score 0-27; recall period: past 2 wk (2-5 min)	Higher scores indicate more depressive symptoms; cut points 5, 10, and 15 for mild, moderate, and severe depression	IBD = 7.6 (95% CI, 6.3-8.8; n = 926)94	Contents overlap with common symptoms of inflammatory disorders (eg, lack of energy, sleep disturbance, appetite changes)77
Anxiety
PROMIS Anxiety65	8 items, each scored 1-5, total score transformed scores into T-scores; recall period: past wk (2 min)	Higher scores indicate more anxiety symptoms; T-score = 50 and SD = 10 represent general population; T-score > 50 abnormal, 2.5 MCID65, 68	IBD T-score = 52 (SD = 10, n = 10,634); UC = 52 (SD = 10, n = 6689); CD = 52 (SD = 9, n = 3945)69	Developed by the NIH77; validated in patients with IBD and has psychometric properties compatible with HADS-A69, 77
HADS-A76	7 items, each scored 0-3, summed score 0-21; recall period: past 7 d (2 min)	Higher scores indicate more anxiety; cut points 8, 11, and 15 for mild, moderate, and severe anxiety	IBD mean = 7.5 (95% CI, 7.1-7.9; n = 8088)94; CD = 7.0 (95% CI ± 0.85; n = 1818); UC = 6.4 (95% CI ± 0.5; n = 1423)53	Most commonly used in patients with IBD94
STAI80	40 items (20 for trait, 20 for state), each scored 1-4, summed subscore range 20-80; recall period: “in general,” scored as “almost never,” “sometimes,” “often,” or “almost always” (10 min)	Higher scores indicate more anxiety; cut point ≥40 for clinically significant anxiety (state is situation-specific; trait is personality trait)	IBD mean = 43.0 (95% CI, 40.1-45.8; n = 857)94	Second most commonly used scale in patients with IBD 94; copyright and requires purchase for use
GAD-779	7 items, each scored 0-3, summed score 0-21; recall period: past 2 wk (2-5 min)	Higher scores indicate more anxiety; cut points 5, 10, and 15 for mild, moderate, and severe anxiety	Median 3, interquartile range 1-7 (n = 242)77	Commonly used in primary care setting but rarely in IBD populations57; lower sensitivity in IBD patients as compared to HADS-A and PROMIS Anxiety77
Stress
PROMIS Stress65	8 items, each scored 1-5, total score transformed scores into T-scores; recall period: past wk (2 min)	Higher scores indicate more stress; T-score = 50 and SD = 10 represent general population; T-score > 50 abnormal, 2.5 MCID65, 68	Not established	Developed by the NIH but not tested in the IBD population
PSS-1072	10 items, each scored 0-4, summed score 0-40; recall period: past 4 wk (3 min)	Higher scores indicate more perceived stress; cutoffs 13 and 27 for moderate and high stress	IBD mean = 20.3 (SD = 7.2, n = 134)95	A 14-item PSS has been used in prior studies of patients with IBD55 but has weaker psychometric properties than the 10-item version75, 95
Fatigue
PROMIS Short Form Fatigue65	8 items, each scored 1-5, total score transformed scores into T-scores; recall period: past wk (2 min)	Higher scores indicate more fatigue; T-score > 50 abnormal, 2.5 MCID68	IBD T-score = 56 (SD = 11, n = 10,634); UC = 56 (SD = 11, n = 6689); CD = 54 (SD = 11, n = 3945)69	Developed by the NIH and has compatible psychometric performance to the PSQI70; validated in IBD69; infrequently used in studies7, 86, 90
FACIT-F84	13 items, each scored 0-4, summed score 0-52; recall period: past 7 d (3 min)	Lower scores indicate more fatigue; cutoff ≤ 30 for clinically significant fatigue; ≥4 point increase indicates clinically significant improvement5	IBD = 38.9 ± 11.0 (n = 209); CD = 38.6 ± 11.3 (n = 111); UC = 39.4 ± 10.6 (n = 96)82	Most widely used; validated in patients with IBD and correlates with serum inflammatory markers5, 82
MFI84	20 items, each scored 1-5, for 5 dimensions of fatigue (4 questions each with dimensional summed score 4-20); recall period: “lately” (5-10 min)	Higher scores indicate greater levels of fatigue; cutoff of ≥13 indicates significant fatigue	Not established	Multifactorial description of fatigue (general, physical, mental, activity, motivation)84; widely used and validated in chronic diseases but not in IBD5, 40
IBD-F scale56	30 items: 25 items, each scored 0-4, subsection scores 0-20 (quantifies fatigue) and 0-120 (impact of fatigue); 5 unscored open-ended questions; recall period: past 30 d (10 min)	Higher scores indicate higher fatigue symptoms and impact	Not established	Developed in consultation with patients with IBD and designed as an IBD-specific fatigue scale but has not been widely used5, 57
Sleep Quality
PROMIS Sleep Disturbance 8a Short Form70	8 items, each scored 1-5, total score transformed scores into T-scores; recall period: past wk (2 min)	Higher scores indicate more sleep disturbance; T-score = 50 and SD = 10 represent general population; T-score > 50 abnormal, 2.5 MCID	IBD T-score = 52 (SD = 9, n = 10,634); UC = 53 (SD = 9, n = 6689); CD = 52 (SD = 8, n = 3945)69	Developed by the NIH and has psychometric performance compatible with the PSQI70; validated in IBD69; infrequently used in studies7, 86, 90
PSQI89	19 items, each scored 0-3, summed score 0-21 (7 domains: sleep duration, night awakenings, sleep latency, daytime dysfunction, sleep efficiency, sleep quality, use of sleep medications); recall period: past 30 d; (5-10 min)	Higher scores indicate worse sleep quality; >5 points clinically significant sleep disturbance	CD: mean = 7.13 (SD = 4.33, n = 405); UC: mean = 6.36 (SD = 4.22, n = 255)6	Most commonly used in patients with IBD7, 90
ESS91	8 items, each scored 0-3, summed score 0-24; recall period: “in recent times” (2 min)	Higher scores indicate more daytime sleepiness; cutoffs 11, 13, and 16 for mild, moderate, and severe excessive daytime sleepiness	Not established	Used in a few observational studies in patients with IBD in conjunction with PSQI,88 but less correlated with psychological symptoms and sleep diaries than PSQI92

BDI-II indicates the Beck Depression Inventory-II; CD, Crohn disease; CI, confidence interval; ESS, Epworth Sleepiness Scale; FACIT-F, Functional Assessment of Chronic Illness Therapy-Fatigue; GAD-7, Generalized Anxiety Disorder-7; HADS, Hospital Anxiety and Depression Scale; IBD-F, Inflammatory Bowel Disease-Fatigue; MFI, Multidimensional Fatigue Inventory; NIH, National Institutes of Health; PROMIS, Patient-Reported Outcomes Measurement Information System; PHQ-9, Patient-Health Questionnaire-9; PSQI, Pittsburgh Sleep Quality Index; PSS-10, 10-Item Perceived Stress Scale; STAI, State and Trait Anxiety Inventory.

Stress

Early-life stress has been found to be a risk factor in the later development of IBD, and ongoing stressors exacerbate inflammation and relapse.¹² Psychosocial stress has a mediating role in circulating levels of inflammatory cytokines, hypothalamic-pituitary axis, intestinal permeability, and alterations in the fecal and gut mucosal microbiota.^{12, 16} The Perceived Stress Scale (PSS-10) is a standardized, valid, and reliable psychometric measurement of perceived stress in psychophysical health research and among patients with IBD.^{2, 55, 72} The PSS-10 contains 10 items (each scored from 0 to 4, summed score range 0 to 40), with higher scores corresponding to higher levels of perceived stress (Table 2).⁷³ Sample items in the PSS-10 are:

In the last month, how often have you been upset because of something that happened unexpectedly?

In the last month, how often have you felt nervous and “stressed?”

The response scale is as follows: 0 = never, 1 = almost never, 2 = sometimes, 3 = fairly often, and 4 = very often.

Anxiety and Depression

Patients with IBD, particularly those with symptomatically active disease, have higher rates of both anxiety and depression as compared to healthy control patients and individuals with chronic health conditions.⁵³ Diagnoses of depression and anxiety may precede IBD diagnoses by as much as 3 to 5 years, based on data from 3 large studies in North America and the United Kingdom (the Nurses’ Health Study in the United States,⁷⁴ a primary care database in the United Kingdom,⁴ and a population-based administrative provincial health database in Manitoba, Canada⁷⁵). However, longitudinal prospective studies are lacking to support the causal relationship between depression/anxiety and IBD.¹²

The Hospital Anxiety and Depression Scale (HADS)⁷⁶ is one of the most commonly used valid and reliable instruments to screen for anxiety and depression symptoms in studies of patients with IBD.^{12, 42, 53, 77} It contains 14 items (each scored from 0 to 3; summed subscale ranges from 0 to 21 for depression and anxiety), with higher scores indicating more anxiety or depression symptoms (Table 2).⁷⁶ Sample items from HADS are as follows:

I feel life is not worth living.

Worrying thoughts constantly go through my mind.

I feel restless as if I have to be on the move.

The response scale is as follows: 0 = no, not at all; 1 = no, not much; 2 = yes, sometimes; and 3 = yes, definitely.

Other commonly used validated instruments for depression and anxiety are the Patient-Health Questionnaire,⁷⁸ the Generalized Anxiety Disorder-7,⁷⁹ the State and Trait Anxiety Inventory,⁸⁰ and the Beck Depression Inventory-II.⁸¹ However, the content areas in the Patient-Health Questionnaire overlap with common symptoms of inflammatory disorders (eg, lack of energy, sleep disturbance, appetite changes) and the Beck Depression Inventory-II can overestimate depression in medically ill populations because of its emphasis on somatic symptoms (eg, weight loss, appetite).⁵³ In comparison, the HADS minimizes the overlapping of physical symptoms with depressive/anxiety symptoms and requires one-quarter of the time for patients to complete the questions as compared to the other instruments.^{36, 42, 53} Both the HADS and the State and Trait Anxiety Inventory are protected by copyright, and there is a licensing fee for their authorized use.

Fatigue

Fatigue is a multidimensional symptom with physiologic, emotional, behavioral, and cognitive components that have been defined as “an overwhelming, sustained sense of exhaustion and decreased capacity for physical and mental work.”⁸² Fatigue is one of the most common disabling symptoms reported by more than 80% of adults with IBD.⁵

The Functional Assessment of Chronic Illness Therapy-Fatigue (FACIT-F) is the most commonly used valid and reliable instrument to measure fatigue in studies of patients with IBD.^{40, 83} Although originally developed for measuring fatigue in patients with anemia receiving oncology care, the FACIT-F has been validated in U.S. patients with IBD and has shown high internal consistency, reliability, and construct validity with strong correlations with symptomatic disease activity (using the Harvey-Bradshaw Index for Crohn disease and the Simple Clinical Colitis Activity Index for UC) and with biochemical surrogate markers of disease activity (C-reactive protein, erythrocyte sedimentation rate, and hematocrit).⁸² The FACIT-F contains 13 items (each scored from 0 to 4, summed range 0 to 52), with higher scores indicating better functioning and less fatigue (Table 2).⁸² Sample items from FACIT-F are as follows:

I have trouble finishing things because I am tired.

I have to limit my social activity because I am tired.

The response scale is as follows: 0 = not at all, 1 = a little bit, 2 = somewhat, 3 = quite a bit, and 4 = very much.

The Multidimensional Fatigue Inventory (20 items, each scored from 1 to 7)⁸⁴ is another commonly used scale to assess for fatigue, but it has not been validated for use in patients with IBD.⁵⁶ Another less commonly used measurement is the IBD-Fatigue scale (25 items scored from 0 to 4, and 5 unscored open-ended questions), which was developed with input from patients with IBD but has not been widely used (Table 2).^{38, 56}

Sleep Disturbance

Sleep quality impairments are common in patients with IBD, even during clinical remission (up to 77%).¹⁰ Sleep impairment can activate inflammatory pathways and has been found to be associated with histologic inflammation,⁸⁵ endoscopic inflammation,⁸⁵ risk of Crohn disease relapse,⁸⁶ hospitalization,¹⁰ and surgery.¹⁰ Moreover, sleep disturbances and excessive daytime sleepiness in patients with IBD may be driven by factors beyond the measure of objective sleep parameters such as polysomnography,⁸⁷ actigraphy (motion-based watch-like devices that record sleep patterns and circadian rhythms),⁸⁸ and melatonin metabolites.⁸⁸

The Pittsburgh Sleep Quality Index (PSQI)⁸⁹ is the most commonly used valid and reliable assessment tool in research studies in IBD.^{6, 7, 90} The PSQI contains 19 items (each scored from 0 to 3, summed score range 0 to 21), with higher scores indicating more sleep dysfunction (in terms of sleep duration, disturbance, latency, efficiency, and quality; medication use; and daytime sleepiness; Table 2).⁸⁹ Sample items from the PSQI are as follows:

During the past month, how often have you taken medicine (prescribed or over the counter) to help you sleep?

If you have a roommate or bed partner, ask him/her how often in the past month you have had loud snoring.

The response scale is as follows: 0 = not during the past month, 1 = less than once a week, 2 = once or twice a week, and 3 = three or more times a week.

An alternative validated PRO instrument for sleep quality that has been used in patients with IBD^{5, 34, 88} is the Epworth Sleepiness Scale (8 items, each scored from 0 to 3, summed range 0 to 24), which measures daytime sleepiness (Table 2).⁹¹ However, the Epworth Sleepiness Scale is less correlated with psychological symptoms and sleep diary assessments as compared to the PSQI.⁹²

DATA ANALYSIS

Assessment of Endpoints

In general, PRO measures are ordinal or continuous in nature and endpoints for determining treatment efficacy can be framed as (1) a continuous variable, a comparison of means to estimate the difference in treatment effects between groups, or (2) a dichotomous variable, a responder analysis of the proportion of patients within each arm who reach a predefined threshold that “constitutes clinically meaningful within-patient change” (Figure 1).^{24, 30, 61} A dichotomous responder analysis that compares the proportion of patients in each treatment/control arm who experience an improvement at least as large as the MCID is less susceptible to the effects of highly skewed outliers than a comparison of means.^{61, 62} The FDA does not recommend the analysis of endpoints from a percentage change from baseline unless “the target response is complete resolution of signs and symptoms” because of poor statistical measurement properties (eg, floor and ceiling effects, higher symptom burden for trial entry).³⁰

An example of 1 study that analyzed endpoints as continuous and dichotomous variables is summarized as follows: A parallel-group multicenter RCT evaluated the impact of eight 1-hour weekly CBT sessions on an intervention group (n = 59) compared to a waitlist control group (n = 59) on anxiety and depressive symptoms.⁴³ In an intention-to-treat analysis (with multiple imputation using chained equations for missing data), outcomes were assessed as a continuous variable for the change in HADS scores from baseline to follow-up at 3.5 months and were compared between the intervention group and the control group using analyses of covariance and reported as the difference and its 95% confidence interval. In a secondary analysis, the proportion of patients with HADS scores >10 (cutoff for moderate to severe anxiety/depressive symptoms⁷⁶) was a dichotomous variable that was compared between the intervention and control groups at baseline and at follow-up, reported as effect size and its 95% confidence interval.

Time-to-event endpoints can also represent clinical outcomes of interests in interventional studies in patients with IBD, such as time to disease relapse/flare, escalation of IBD therapy, emergency department visits, hospitalizations, surgery, and death.³³

Intention-to-Treat, per-Protocol, and Complete-Case Data Analysis

Intention-to-treat analysis, where baseline and outcomes data are analyzed for “all patients randomized according to assigned treatment arm, regardless of adherence,” is most often preferred to preserve the benefits of randomization and minimize bias.^{28, 93} Strict intention-to-treat analysis is often difficult to achieve in pragmatic clinical trials, primarily because of nonadherence to the trial protocol (eg, withdrawal from the study after enrollment/randomization, failure to attend psychotherapy sessions) and missing outcomes (eg, no-show at follow-up visits, nonresponse on assessment surveys).⁹³ The imputation of missing data is commonly performed to conform to intention-to-treat analysis,⁹³ and examples from 2 recently published RCTs of psychotherapeutic interventions in patients with IBD include a simple imputation method of replacing the missing final outcome variable with the last known value before the participant was lost to follow-up (“last observation carried forward”)⁴⁴ and a more complex multiple imputation using chained equations (pooled estimates from multiple imputed data sets).⁴³

Nonadherence to the assigned intervention may underestimate its potential benefit in the intention-to-treat analysis.⁹³ The FDA accepts secondary analysis using per-protocol and complete-case protocol (nonrandomized observational comparisons) as sensitivity analysis to determine the impact of missing data,²⁸ which can be common in RCTs for voluntary behavioral interventions for psychological comorbidities in IBD.^{34, 44, 60} Complete-case analysis includes the outcomes of all participants who have follow-up assessments, whereas per-protocol analysis includes only the participants who fulfill the trial protocol and outcome assessments.⁹³

A recently published RCT illustrates the various data analysis methods discussed above. This RCT investigated whether 8 weekly acceptance and commitment therapy sessions reduced stress in patients with clinical-biochemical quiescent IBD as compared to control patients (who received treatment as usual).⁴⁴ A total of 122 patients were enrolled and were randomized 1:1 to the intervention (n = 61) and control (n = 61) groups. After enrollment and randomization, 21 patients withdrew from the study and 22 failed to attend 1 of the follow-up visits. The primary analysis was complete-case intention-to-treat analysis that included 79 participants for whom data were available at weeks 0 (baseline), 8 (end of treatment), and 20 (follow-up). An exploratory analysis with 101 participants with partial or full follow-up data was performed using the last-observation-carried-forward to impute missing data for the 22 participants. A per-protocol analysis of psychological variables was also performed for 33 participants who attended at least 5 out of 8 or more acceptance and commitment therapy sessions.⁴⁴

APPROACHES TO MINIMIZE BIAS

Randomization

Randomization, blinding or masking, and control groups can help discriminate between the effects of the intervention effects and the natural history of the disease while minimizing measurement or reporting bias.²³ Types of control groups include placebo, active control, no treatment, and sham.²³ The FDA favors the randomized allocation of study participants as the preferred means of ensuring the comparability of test groups and minimizing the possibility of selection bias in clinical trials.²⁶ Randomization helps reduce bias associated with treatment selection, and although randomization cannot guarantee balance for all participant characteristics between treatment groups, there is an inference of balance for factors that may be unknown or unmeasured.^{43, 59} The stratification of participants for known, measurable confounding variables (eg, IBD disease activity^{36, 37, 45, 53}) can decrease between-treatment group variations and increase the reliability of the results.³² However, overstratification can limit the feasibility of trials because a very large sample size would then be needed for the randomization of participants into each stratum.⁵⁹

Blinding/Masking

Blinding (concealing of participant assignment) can prevent and reduce bias in the reporting or interpretation of psychological or physical responses by study participants, investigators, and/or assessors.^{23, 37} The blinding of study participants is particularly important for PROs because the knowledge of treatment assignment can affect participant behavior and responses, which has been referred to as the Hawthorne effect.⁵¹ Nonetheless, blinding can be challenging, impractical, or unethical in certain types of trials.^{23, 45, 46} In open-label studies (investigators and participants are not blinded to the treatment allocation), the FDA recommends “the use of pre-specified decision rules for aspects of study conduct, such as treatment assignment, patient management, safety reporting, and response variable ascertainment” to reduce the consequences of the lack of blinding.²³ Although it is often not possible to blind participants to their allocated group, blinding investigators can reduce bias at the stage of randomization,⁴⁴ clinical outcome assesments,^{39, 44, 49} and outcome adjudication.⁴⁹

The National Institutes of Health published the Behavioral and Social Science Research (BSSR) Involving Humans clinical trials template in 2019, which can be used as a resource for researchers preparing behavioral-based interventions in humans (https://grants.nih.gov/policy/clinical-trials/protocol-template.htm).

Conclusions

In summary, RCTs with pharmacological and nonpharmacological interventions with primary endpoints assessed by PRO measurements are needed to effectively address neurobiological symptoms prevalent among individuals with IBD, particularly those with active disease. It is possible that a combination of conventional anti-inflammatory therapy for IBD and adjunctive pharmacotherapy/psychological treatments directed at the gut-brain axis can modulate neurobiological symptoms common in patients with IBD, including stress, anxiety, depression, fatigue, and sleep disturbances. The MCID of validated instruments can aid in the interpretation of statistical significance and clinical relevance. Trial designs with randomization, masking, and control groups can reduce bias and increase the reliability of the results.

Conflicts of interest: Dr. Sandborn reports research grants from AbbVie, Abivax, Arena Pharmaceuticals, Boehringer Ingelheim, Celgene, Genentech, Gilead Sciences, GlaxoSmithKline, Janssen, Lilly, Pfizer, Prometheus Biosciences, Seres Therapeutics, Shire, Takeda, and Theravance Biopharma; consulting fees from AbbVie, Abivax, Admirx, Alfasigma, Alimentiv (previously Robarts Clinical Trials, owned by Alimentiv Health Trust), Alivio Therapeutics, Allakos, Amgen, Applied Molecular Transport, Arena Pharmaceuticals, Bausch Health (Salix), Beigene, Bellatrix Pharmaceuticals, Boehringer Ingelheim, Boston Pharmaceuticals, Bristol Myers Squibb, Celgene, Celltrion, Cellularity, Cosmo Pharmaceuticals, Escalier Biosciences, Equillium, Forbion, Genentech/Roche, Gilead Sciences, Glenmark Pharmaceuticals, Gossamer Bio, Immunic (Vital Therapies), Index Pharmaceuticals, Intact Therapeutics, Janssen, Kyverna Therapeutics, Landos Biopharma, Lilly, Oppilan Pharma, Otsuka, Pandion Therapeutics, Pfizer, Progenity, Prometheus Biosciences, Prometheus Laboratories, Protagonists Therapeutics, Provention Bio, Reistone Biopharma, Seres Therapeutics, Shanghai Pharma Biotherapeutics, Shire, Shoreline Biosciences, Sublimity Therapeutics, Surrozen, Takeda, Theravance Biopharma, Thetis Pharmaceuticals, Tillotts Pharma, UCB, Vendata Biosciences, Ventyx Biosciences, Vimalan Biosciences, Vivelix Pharmaceuticals, Vivreon Biosciences, and Zealand Pharma; stock or stock options from Allakos, BeiGene, Gossamer Bio, Oppilan Pharma, Prometheus Biosciences, Prometheus Laboratories Progenity, Shoreline Biosciences, Ventyx Biosciences, Vimalan Biosciences, and Vivreon Biosciences; and is an employee at Shoreline Biosciences. Dr. Singh is supported by the National Institutes of Health/National Institute of Diabetes and Digestive and Kidney Diseases. Dr. Tse has no conflicts to disclose.