Minimal-Contact Versus Standard Cognitive Behavioral Therapy for Irritable Bowel Syndrome: Cost-Effectiveness Results of a Multisite Trial

Abstract

Background

Irritable bowel syndrome (IBS) is a common, often disabling gastrointestinal (GI) disorder for which there is no satisfactory medical treatment but is responsive to cognitive behavior therapy (CBT).

Purpose

To evaluate the costs and cost-effectiveness of a minimal contact version of CBT (MC-CBT) condition for N = 145 for IBS relative to a standard, clinic-based CBT (S-CBT; N = 146) and a nonspecific comparator emphasizing education/support (EDU; N = 145).

Method

We estimated the per-patient cost of each treatment condition using an activity-based costing approach that allowed us to identify and estimate costs for specific components of each intervention as well as the overall total costs. Using simple means analysis and multiple regression models, we estimated the incremental effectiveness of MC-CBT relative to S-CBT and EDU. We then evaluated the cost-effectiveness of MC-CBT relative to these alternatives for selected outcomes at immediate posttreatment and 6 months posttreatment, using both an intent-to-treatment and per-protocol methodology. Key outcomes included scores on the Clinical Global Impressions-Improvement Scale and the percentage of patients who positively responded to treatment.

Results

The average per-patient cost of delivering MC-CBT was $348, which was significantly less than the cost of S-CBT ($644) and EDU ($457) (p < .01). Furthermore, MC-CBT produced better average patient outcomes at immediate and 6 months posttreatment relative to S-CBT and EDU (p < .01). The current findings indicated that MC-CBT is a cost-effective option relative to S-CBT and EDU.

Conclusion

As predicted, MC-CBT was delivered at a lower cost per patient than S-CBT and performed better over time on the primary outcome of global IBS symptom improvement.

A largely home-based regimen of cognitive behavioral therapy for irritable bowel syndrome was delivered at a lower cost per patient than traditional office-based CBT and performed better over time in relieving core gastrointestinal symptoms (e.g., abdominal pain, altered bowel habits such as diarrhea and/or constipation) refractory to conventional medical treatments.

Introduction

Irritable bowel syndrome (IBS) is a functional gastrointestinal (GI) disorder (also known as disorder of brain–gut interaction) characterized by abdominal pain associated with diarrhea and/or constipation. It is one of the most common GI diseases diagnosed by gastroenterologists and one of the most common disorders seen by primary care physicians [1]. Lacking a reliable biomarker that corresponds with symptom severity, IBS is best understood from the perspective of a biopsychosocial model [2, 3] defined by individual biology (e.g., altered GI physiology, abnormal central processing of noxious stimuli, altered autonomic nervous system), behavior (e.g., excessive avoidance, reassurance seeking and other safety behaviors), environment (e.g., modeling, reinforcement contingencies, interpersonal difficulties), emotion (e.g., negative affect), and higher-order cognitive processes (e.g., restricted coping, hypervigilance, cognitive biases, selective attention to threat cues). Their interactive and reciprocal influence on symptom perception appears particularly pronounced in individuals with psychological (e.g., neuroticism, disruptive early life experiences such as trauma, parenting practices) and/or biological (genetics, prior GI infection, intestinal microbiota) vulnerabilities that heighten reactivity (emotional, physiological) under stressful life circumstances perceived as uncontrollable, unpredictable, and/or overwhelming. The combination of these factors can lead to alterations in the structure and function of several brain networks (including those governing salience detection, emotional arousal, and sensorimotor processing) [4] and in bidirectional brain–gut interactions, resulting in characteristic symptoms [3]. Their unresponsiveness to conventional medical and dietary treatments contributes to direct (healthcare) economic costs estimated conservatively at $1 billion [5] in the USA alone and quality of life impairment comparable to other chronic diseases, such as diabetes mellitus and hepatitis [6, 7].

Given its economic and personal burden, there is a need for behavioral self-management treatments that not only provide relief of IBS refractory clinical symptoms but are cost effective and, particularly in the COVID era, practical (i.e., accessible, less resource-intensive) and preferable to patients, providers, and health care systems struggling with limited resources. Of treatments of any orientation (pharmacological, behavioral, dietary), one with a particularly strong efficacy profile is cognitive behavioral therapy (CBT) [8]. The National Institute for Health and Care Excellence (NICE) [9], an independent body responsible for improving the quality of the English health care system, recommends CBT for patients with refractory IBS symptoms (i.e., ongoing symptoms after 12 months despite being offered appropriate medications and therapeutic lifestyle change). As part of its effort to promote evidence-based practice in the USA patterned after NICE guidance, the Society of Clinical Psychology (Division 12) of the American Psychological Association [10] has identified a specific regimen of CBT [11, 12] an empirically validated treatment for IBS, a distinction assigned to a select number of psychological treatments deemed superior to a psychological placebo, drug, or other treatments based on evidence from multiple controlled studies involving at least two investigative teams [10].

Favorable conclusions regarding the therapeutic benefit of CBT were reinforced by the results of a pivotal National Institutes of Health (NIH)-funded multisite trial called the IBS Outcome Study [13] that pitted a largely home-based CBT with minimal therapist contact (MC-CBT) against the standard, clinic-based CBT (S-CBT) delivered one-on-one in an outpatient setting across successive weeks with reference to a nonspecific education/support comparator (EDU) that controlled for therapeutic factors (e.g., time, attention, quality of therapeutic alliance) common to any given treatment. MC-CBT produced a statistically significant (p < .05) larger percent of treatment responders than EDU at immediate posttreatment whether rated by patients (61.0% vs. 43.5%) or study gastroenterologists (GE) blind to treatment assignment (55.7% vs. 40.4%) 2 weeks after treatment completion [12]. IBS symptom improvement rates posttreatment are among the highest in the IBS outcome literature (>100 RCTs) when examined on an absolute level. To put these data in context, the proportion of treatment responders of two FDA-approved pharmacological agents using global IBS symptom improvement scales have been 17% and 40% [14, 15], respectively.

A primarily home-based version of CBT that requires fewer provider resources should be less costly to deliver from the provider perspective than a clinic-based version of CBT; however, its economic impact in terms of costs combined with effectiveness (i.e., cost-effectiveness) is not known. Beyond clarifying its efficacy profile, cost-effectiveness data equips health policymakers with information on which to base decisions about the allocation of resources to help reduce the burden of suffering attributable to IBS. Unfortunately, few trials have prospectively examined the costs and cost-effectiveness of behavioral therapies for IBS, however, two studies [16, 17] did examine the cost-effectiveness of behavioral therapies with or without medications. Andersson et al. [16] examined the cost-effectiveness of internet-delivered CBT for IBS relative to a discussion forum. They found that the internet-delivered CBT produced cost reductions for the intervention group relative to the discussion forum group with an incremental cost-effectiveness ratio of $16,806. Furthermore, 36% of the participants receiving the internet-delivered CBT were deemed recovered based on scores from the Gastrointestinal Symptom Rating Scale-IBS version (GSRS-IBS) compared to only 2% of those participants in the forum discussion group, and this represented a 15-fold increase in the expected rate of recovery in the intervention group relative to the discussion forum group. In another study, Creed et al. [17] examined the cost-effectiveness of interpersonal psychodynamic therapy relative to the paroxetine and usual care (patients receiving whatever management was deemed appropriate by their regular seen by a gastroenterologist or general practitioner) for severe IBS. They found that both psychotherapy and paroxetine outperformed usual care in improving physical health based on scores from the physical health component of the SF-36 (36-Item Short Form Survey) questionnaire. The change in scores from baseline to 15 months was 5.2 for psychotherapy, 5.8 for paroxetine, and –0.3 for usual care (ANCOVA p < 0.001). However, they found no difference in psychological health across the three conditions and their choice of psychotherapy was no more effective than usual care in IBS symptoms (i.e., pain).

The present economic analysis was part of a larger study designed to determine the 6-month effectiveness and cost-effectiveness of minimal contact CBT relative to the “gold standard” version delivered over 10 consecutive weeks by clinicians in an outpatient setting and a 4-session IBS education intervention. Specifically, Lackner et al. [12] randomly assigned 436 adult patients seeking treatment for Rome-diagnosed IBS at one of two large, geographically distinct university-affiliated sites in the northern USA to one of three treatment conditions—minimal contact CBT [MC-CBT, N =145], standard CBT [S-CBT, N = 146], or an educational group [EDU], N = 145). The EDU condition was included as a nonspecific condition controlling for time, attention, etc. and emphasized supportive counseling and education allowing us to isolate treatment benefit due to CBT’s technical procedures [12]. Patients were enrolled between 2010 and 2015 (inclusive), and the study achieved a sample size sufficient to achieve the power of 0.90 for an adjusted mean difference of d = 0.59 (using Cohen’s d) [12].

In this study, we present findings from the cost and cost-effectiveness analysis for the Lackner et al. study. First, we estimated the costs of delivering each of the interventions. Second, we estimated the incremental cost, incremental effectiveness, and incremental cost-effectiveness ratio by comparing the difference in cost and outcomes, if any, between patients who were randomly assigned to MC-CBT and those who received S-CBT or EDU. The effectiveness measures used for the economic study were developed for the study’s main findings paper [12] and included two measures based on patient responses to the Clinical Global Impressions-Improvement Scale (CGI-I) [18], which provides a measure of overall IBS symptom improvement and two additional outcome measures based on physician-reported CGI that paralleled the patient-reported measures.

Methods

Study Design

The study design and eligibility criteria have been described previously by Lackner et al. [13]. Adult patients (18–70 years) seeking treatment for Rome-diagnosed IBS were randomized into three treatment conditions—MC-CBT, S-CBT, or EDU. CBT targets both dysfunctional patterns of thinking (intense worry, rigid problem solving) and behaviors (e.g., stress reactivity) believed to maintain IBS. Patients in the MC-CBT condition received up to four 60-min sessions over 10 weeks supplemented by home study materials. In addition, these patients received up to two 10-min phone contacts in weeks 3 and 7 of treatment to troubleshoot around home exercises. Patients in the S-CBT condition received treatment once-a-week 60-min sessions over a 10-week acute phase. The EDU condition provided information about IBS, its clinical features, causes, treatments, and the role of lifestyle factors (stress, diet, exercise) in the context of a supportive therapeutic environment without prescribing behavior changes nor addressing any CBT mechanisms. EDU was delivered in four 60-min sessions over a 10-week acute phase. All sessions were protocol-driven and conducted by competency-trained PhD level psychologists with different allegiances (CBT, hypnotherapy, psychodynamic therapy) reflective of real-world realities of clinical practice. For a detailed description of the treatments, see [13]. The economic analysis was designed as an integral part of this study, and the data collection instruments used were developed in collaboration with the economics team as part of a 1-year planning phase to ensure that data captured would support the economic aims. Table 1 presents baseline data for the sample.

Table 1.

Baseline sociodemographic and clinical characteristics by treatment condition

Characteristic	Overall (n = 436)	MC-CBT (n = 145)	S-CBT (n = 146)	EDU (n = 145)
Age, mean (SD)	41.4 (14.8)	40.9 (14.6)	41.1 (14.4)	42.2 (15.4)
Women, N (%)	350 (80.3%)	124 (85.5%)	112 (76.7%)	114 (79.2%)
Race/ethnicity, N (%)
Non-Hispanic White	390 (89.4%)	133 (91.7%)	128 (87.7%)	129 (89.0%)
African-American	28 (6.4%)	8 (5.5%)	9 (6.2%)	11 (7.6%)
Other or missing	18 (4.2%)	4 (2.8%)	9 (6.2%)	5 (3.5%)
Marital status, N (%)
Never married	185 (42.4%)	61 (44.1%)	60 (41.1%)	64 (44.1%)
Married	185 (42.4%)	68 (46.9%)	58 (39.7%)	59 (40.7%)
Separated/divorced	57 (13.1%)	11 (7.6%)	26 (17.8%)	20 (13.8%)
Widowed	9 (2.1%)	5 (3.4%)	2 (1.4%)	2 (1.4%)
Income ($), mean (SD)	74.0 (54.2)	77.9 (56.4)	73.1 (52.2)	71.3 (54.0)
Education, N (%)
High school or less	99 (22.7%)	31 (21.4%)	30 (20.5%)	38 (26.2%)
Associate or Vo-tech	65 (14.9%)	25 (17.2%)	22 (15.1%)	18 (12.4%)
College degree	142 (32.6%)	54 (37.2%)	41 (28.1%)	47 (33.1%)
Post-grad degree	127 (29.1%)	35 (24.1%)	52 (35.6%)	40 (27.6%)
Missing	3 (0.7%)	0	1 (0.7%)	2 (1.4%)
Employment status, N (%)
Employed full- or part-time	277 (63.5%)	92 (63.4%)	91 (62.3%)	94 (64.8%)
Unemployed	109 (25.0%)	38 (26.2%)	40 (27.4%)	31 (21.4%)
Homemaker	13 (3.0%)	4 (2.8%)	5 (3.4%)	4 (2.8%)
Retired	33 (7.6%)	9 (6.2%)	9 (6.2%)	15 (10.3%)
Missing	4 (0.9%)	2 (1.4%)	1 (0.7%)	1 (0.7%)
Predominant bowel type, N (%)
Constipation	130 (29.8%)	43 (29.7%)	40 (27.4%)	47 (32.4%)
Diarrhea	188 (43.1%)	59 (40.7%)	67 (45.9%)	62 (42.8%)
Mixed	98 (22.5%)	33 (22.8%)	35 (24.0%)	30 (20.7%)
Undifferentiated	20 (4.6%)	10 (6.9%)	4 (2.7%)	6 (4.1%)
Years with IBS, mean (SD)	17.1 (14.4)	15.7 (13.3)	17.7 (13.3)	17.7 (16.4)
Received medical care for IBS (lifetime), N (%)	328 (75.2%)	107 (73.8%)	116 (79.5%)	105 (72.4%)
IBS treatment-naïve, N (%)	10 (2.2%)	4 (2.6%)	3 (1.9%)	3 (1.9%)
Assessment scores, mean (SD)
IBS Symptom Severity Scalea	281.9 (72.1)	278.0 (68.6)	285.1 (76.7)	282.4 (71.0)
Brief Symptom Inventorya
Anxiety	4.50 (4.50)	4.22 (4.26)	4.27 (4.41)	5.02 (4.81)
Depression	3.97 (4.29)	4.07 (4.47)	3.82 (4.33)	4.03 (4.09)
Somatization	4.22 (3.93)	4.16 (4.31)	4.00 (3.56)	4.54 (3.91)
Global Severity Index	12.7 (11.0)	12.4 (11.6)	12.1 (10.5)	13.6 (10.8)
Medical comorbidities, #	4.6 (4.9)	4.8 (5.2)	4.3 (4.7)	4.8 (5.0)
Psychiatric comorbidities, #	1.2 (1.6)	1.1 (1.5)	1.3 (1.7)	1.2 (1.7)
Medication use for IBS symptoms (N, %)	292 (67.0%)	94 (64.8%)	95 (65.1%)	103 (71.0%)
Pain medication	35 (8.0%)	9 (6.2%)	13 (8.9%)	13 (9.0%)
Bowel medication	271 (62.2%)	86 (59.3%)	87 (59.6%)	98 (67.6%)
Multisymptom medication	20 (4.6%)	6 (4.1%)	7 (4.8%)	7 (4.8%)
Psychiatric medication	26 (6.0%)	8 (5.5%)	12 (8.2%)	6 (4.1%)

Note. ^aHigher scores indicate more severe symptoms; IBS-SSS ≥300 = severe.

Cost Analysis

We used a microcosting approach to compute the service-level costs of the treatment conditions from the provider perspective. Costs were calculated from the provider perspective as this perspective is the most relevant for providers looking to implement these types of therapies in real-world practice. Therefore, we only included the labor time and other nonlabor costs incurred by the provider sites to deliver these services to the patients. In this analysis, we did not include the value of the patient’s time or other costs that he/she may have incurred to undergo this intervention (e.g., travel costs to/from site). In addition, because this was a research study, we wanted to estimate treatment costs that would be incurred in real-world settings (as opposed to those required to implement a clinical trial research protocol). Therefore, we identified the clinical activities and services that constituted each intervention. We did not include nontreatment activities related to implementing the clinical research trial such as randomization and conducting follow-up assessments because these costs would not be incurred in real-world clinical practice. As described below, using a service-level costing approach allowed us to get detailed data on the resources used (and associated unit prices) to provide specific services and perform related activities in the delivery of the interventions. These data were collected prospectively for each patient with quality checks performed at multiple points to ensure robust cost data. All costs were converted to 2016 U.S. dollars using the Bureau of Labor Statistics’ Consumer Price Index (http://www.bls.gov/cpi/). Statistical tests (t-tests) were performed to determine whether estimated means for time and intervention costs were statistically significantly different between the alternative interventions.

Labor Costs

For each patient, attendance was tracked to capture the number of services received by the patient and the duration of each service. This information was summed across services to calculate the total direct treatment time each patient received within each treatment condition. Study intervention staff also estimated the average indirect time (e.g., session preparation, note-taking, administrative work) associated with each type of service. Again, by summing across services, we calculated the total indirect treatment time for each patient within each intervention condition. Information on therapists’ wages was provided from the study’s principal investigator based on information from financial records and adjusted to 2016 dollars. This wage was estimated to be $55.52 per hour. For each patient, labor costs were calculated by taking the average therapist hourly wage and multiplying by the total time reported (direct plus indirect time) for each session received by the patient.

Nonlabor Costs

Estimates for the building space size and associated costs used for the therapy sessions were provided by the administrative office of the clinics where the study was implemented. Materials costs varied across the treatment conditions due to differences in the materials provided. Each participant who engaged in treatment received the relevant materials packet for that condition. The costs of these materials were estimated to be $19.53, $4.45, and $107.13 (2016$) for MC-CBT, S-CBT, and EDU, respectively. The materials for EDU were most costly due to the provision of IBS-related books, daily diaries, printed self-assessment stress profiles and worksheets. Materials for MC-CBT included a workbook, and relaxation CD. Materials for S-CBT included worksheets and a relaxation CD. The cost of building space (estimated as 1 cent per square foot per hour) was derived using data collected from real estate sources (e.g., Loopnet.com) for the geographic areas in which the interventions were being implemented. Building space and materials costs were added to the patient-level costs, as appropriate, based on the number of sessions received.

Total Costs and Cost per Patient

The total cost of service was simply the sum of costs for labor time, building space, and materials (if applicable). Summing across all services received by the patient yielded the total cost of the treatment for that patient, and summing across all patients in that treatment group yielded the total cost of the treatment condition (i.e., MC-CBT, S-CBT, EDU). We derived the average costs per patient for each condition by dividing the total costs of the condition by the number of patients assigned to that condition.

Effectiveness Measures

For the cost-effectiveness analysis, we used four of the effective measures that were developed for the study’s main findings paper [12] consistent with recommendations for efficacy assessment for functional GI disorders trials [20]. Specifically, two of the efficacy measures were based on patient responses to the Clinical Global Impressions-Improvement Scale (CGI-I) [18], which provides a measure of overall IBS symptom improvement. The relevant question asks: “Compared to how you felt prior to entering the study, how would you rate the IBS symptoms for which you sought treatment during the past week?” (1 = substantially worse, 2 = moderately worse 3, slightly worse, 4= no change, 5 = slightly improved, 6 = moderately improved, 7 = substantially improved). For one outcome, we examined the mean response score in the patient-reported CGI across the three treatment conditions. For a second outcome measure, we calculated the percentage of patients who responded to treatment. To do this, patients whose symptoms were rated as “substantially improved” or “moderately improved” based on data from the CGI were classified as treatment responders. In addition, study gastroenterologists completed a physician-version of the CGI. Therefore, we also created two additional outcome measures based on physician-reported CGI that paralleled the patient-reported measures. For this study, we used outcomes that were measured at immediate posttreatment (approximately 2 weeks from treatment completion) and at 6 months posttreatment. Including both the immediate and 6-month outcomes let us examine the cost-effectiveness initially achieved and whether our findings were sustained 6 months later. As described in the main findings [12] modified linear probability models with Huber–White robust estimators was used for between group-comparisons of the dichotomous outcome variables; and single degree of freedom contrasts with Huber–White robust estimators were performed for between-group mean comparisons of the response score outcomes. The models included site as a covariate as well as covariates representing medication status (patient using medication for abdominal pain, bowel symptoms, or for multiple IBS symptoms versus not), and patient ethnicity (White versus non-White). All outcome measures were developed using an intent-to-treat methodology as well as a per-protocol methodology. In this paper, we present the findings from the intent-to-treat methodology only as the results were similar using both methodologies.

Cost-Effectiveness Analysis

The cost-effectiveness analysis compared MC-CBT to S-CBT and EDU by calculating an incremental cost-effectiveness ratio (ICER) for each of the four outcome measures. The first step in the cost-effectiveness analysis was to rank the treatments in increasing order of average per-patient cost for each of the effectiveness measures. The ICER was then computed, defined as the difference in average cost (C) divided by the difference in average effectiveness (E) where the incremental cost-effectiveness ratio (ICER) equals (C_G – C_S)/(E_G – E_S). If one treatment condition is both less expensive and more effective, it strictly dominates the alternative intervention. When neither therapy is strictly dominated, the cost-effectiveness ratio is calculated regardless of statistical significance.

Choosing the optimal treatment option after strictly dominated options are removed depends on a decision maker’s willingness to pay (WTP). WTP refers to the value that a person (e.g., provider) or group of people (e.g., health insurer, health agency) is willing to pay to achieve a given outcome. Economic theory suggests that the optimal intervention is the one with the greatest ICER that is not more than the decision maker’s WTP for an additional unit of the outcome [21]. We calculated a cost-effectiveness acceptability curve (CEAC) to show the probability that a treatment is the cost-effective option as a function of the decision maker’s WTP for each of the clinical outcomes. The CEAC incorporates the inherent variability of the cost and effectiveness estimates and allows us to better capture the variability in our cost-effectiveness analysis in lieu of confidence intervals for the ICERs [22, 23]. The CEACs were calculated using a nonparametric bootstrap method to draw, with replacement, random samples of size N = 436 (equal to the total sample size) of participant cost and effectiveness estimates by treatment arm. We calculated the average cost and effectiveness for each random sample and calculated the associated ICER. We compared this ICER to alternative WTP values to determine if the treatment arm is cost-effective. To calculate the probability that a treatment arm is cost-effective, we repeated this process over 1,000 iterations across a range of WTP values.

Sensitivity Analysis

To assess the sensitivity of our results, we varied our assumptions on the unit price for labor, which was the key driver of our costs. We applied various wages, assuming both higher and lower costs for labor inputs. This allowed us to examine how labor changes, such as a lower-priced job type delivering the intervention, might impact our cost-effectiveness results. We found that our cost-effectiveness results were not sensitive to changes in the cost of labor as the same unit costs were applied to all three conditions for all treatment activities. While the condition-specific sessions differed in length and type (face-to-face or telephonic) of sessions, the effect of changing the associated labor costs changed costs proportionally across conditions. Sensitivity analysis showed that no differential cost effect across the treatment conditions would occur and the cost-effectiveness conclusions remained unchanged.

Results

Characteristics of Service Utilization

Table 2 shows the average characteristics of the intervention services received by patients in each treatment condition, including average total labor time (direct, indirect, and total) as well as the average number of sessions received per patient, and the average total cost per patient. MC-CBT patients received less total time per patient, on average, across all treatment activities (an average of approximately 6.4 h) compared to S-CBT patients (12.3 h) or EDU patients (6.9 h). The much larger observed variance in average total time between the MC-CBT and S-CBT conditions was primarily due to the condition-specific number of face-to-face individual CBT sessions. MC-CBT patients received fewer individual CBT sessions than S-CBT by design which resulted in less therapy time. In total, MC-CBT patients received an average of 3.7 individual face-to-face sessions with each session lasting approximately 60 min and 1.9 brief telephonic sessions with each telephone call lasting approximately 15 min. S-CBT patients received an average of 8.8 individual CBT sessions with each session lasting 60 min. These average number of sessions attended is slightly lower than the protocol design because not all patients across treatment conditions participated in all planned sessions. The MC-CBT condition had better participation rates with 79% of the patients in that condition attending all planned sessions. Only 69% of patients in the S-CBT condition attended all planned sessions. The EDU condition was similarly designed as the MC-CBT condition with the same amount of sessions and session time to be provided per patient, however, participation in the EDU condition was better with 82% of the study patients receiving all six planned sessions. This resulted in slightly greater session time per patient, on average, for the EDU condition relative to MC-CBT.

Table 2.

Mean characteristics of service utilization

Condition	Number of TX session (face-to-face)	Number of TX session (telephonic)	Total Number of TX session	Direct TX time (face-to-face) (min)	Direct TX time (telephonic) (min)	Total time (min)	Total cost per patient ($)
MC-CBT (N = 145)	3.70a	1.88a	5.59a	222.21a	28.24a	383.12a,b	348a,b
	(0.80)	(0.40)	(1.15)	(48.02)	(5.60)	(79.99)	(116)
	[3.57–3.83]	[1.82–1.95]	[5.40–5.77]	[214.32–230.09]	[27.26–29.23]	[369.99–396.25]	[329–366]
S-CBT (N = 146)	8.82a,c	0.00a,c	8.82a,c	528.90a,c	0.00a,c	738.26b,c	644a,c
	(2.37)	(0.00)	(2.37)	(142.16)	(0.00)	(198.43)	(239)
	[8.43–9.20]	[0.00-0.00]	[8.43–9.20]	[505.65–552.16]	[0.00-0.00]	[705.80–770.72]	[606–682]
EDU (N = 145)	3.78c	1.93c	5.71c	226.76c	28.97c	413.79b,c	457b,c
	(0.68)	(0.28)	(0.93)	(40.91)	(4.20)	(66.62)	(129)
	[3,67-3.89]	[1.89–1.98]	[5.56–5.86]	[220.04–233.47]	[28.28–29.66]	[402.86–429.73]	[436–477]

Note. Standard deviations are presented in parentheses. 95% Confidence intervals are presented in brackets.

^aStatistically significant difference between MC-CBT and S-CBT averages at the .001 level.

^bStatistically significant difference between MC-CBT and EDU averages at the .001 level.

^cStatistically significant difference between S-CBT and EDU averages at the .001 level.

Because per-patient time was less for the MC-CBT treatment, patients in the MC-CBT treatment had lower costs per patient than patients in the S-CBT treatment ($348 versus $644; p < .001). This cost difference was driven mostly by the more frequent, on a per-patient basis, CBT sessions provided in S-CBT. Similarly, MC-CBT also had lower average costs than the patients in the EDU group ($348 vs. $457). This cost difference was driven by a slightly higher attendance rate among the patients in the EDU group compared to patients in the MC-CBT. The primary cost driver in each condition was labor costs associated with the therapist’s time in delivery of the interventions (as shown in Table 3).

Table 3.

Mean costs

Condition	Total labor cost per patient ($)	Total nonlabor cost per patient ($)	Total cost per patient ($)
MC-CBT (N = 145)	324a,b	24a,d	348a,d
	(110)	(6)	(116)
	[306–341]	[23–25]	[329–366]
S-CBT (N = 146)	629a,c	16a,c	644a,c
	(234)	(5)	(239)
	[591–666]	[15–16]	[606–682]
EDU (N= 145)	352b,c	105c,d	457c,d
	(106)	(26)	(129)
	[335–369]	[100–109]	[436–477]

Note. Standard deviations are presented in parentheses. 95% Confidence intervals are presented in brackets. Nonlabor cost includes the value of materials and building space used in providing the intervention. Labor cost includes the value of time spent by the therapist providing the intervention.

^aStatistically significant difference between MC-CBT and S-CBT averages at the .001 level.

^bStatistically significant difference between MC-CBT and EDU averages at the .05 level.

^cStatistically significant difference between S-CBT and EDU averages at the .001 level.

^dStatistically significant difference between MC-CBT and EDU average at the .001 level.

Cost-Effectiveness Analysis Results for CGI Scores

Table 4 shows the results of the cost-effectiveness analysis of the CGI scores following the intent to treat methodology used in the main findings paper [12]. Results are shown for immediate posttreatment and 6 months posttreatment for the patient- and physician-reported CGI scores. For each treatment condition, the average cost per patient and average effectiveness is shown. At posttreatment and 6 months posttreatment, MC-CBT economically dominates both S-CBT and EDU as patients in the MC-CBT condition had better outcomes (higher mean scores on the CGI) which were achieved at a lower per-patient cost. Specifically, at immediate posttreatment MC-CBT patients had a higher average CGI score of 5.6 for the patient-reported CGI and 5.5 for the physician reported CGI. At 6 months post-treatment, the average patient-reported CGI score declined slightly to 5.5; however, the average physician-reported CGI score increased to 5.7. In comparison, S-CBT patients had an average CGI score of 5.5 and 5.4, respectively, at immediate posttreatment. These average scores slightly declined at 6-month posttreatment. EDU patients had an average patient-reported and physician-reported CGI scores of 5.3 and 5.2, respectively. Similar to the MC-CBT, at 6 months posttreatment, patients in the EDU condition had a very small decline in the average patient-reported CGI score, but a slight increase in the average physician-reported CGI score. Because the MC-CBT condition was less costly and more effective, it economically dominated both S-CBT and EDU conditions.

Table 4.

Cost-effectiveness analysis, immediate, and 6 months

Condition	Mean total cost per patient ($, 2016) (SE)	Patient-reported CGI scores				Physician-reported CGI scores
		Immediate posttreatment		6 months posttreatment:		Immediate posttreatment		6 months posttreatment
		Mean effectivenessa (SE)	ICER (ΔC/ΔE, $)	Mean effectivenessa (SE)	ICER (ΔC/ΔE, $)	Mean effectivenessa (SE)	ICER (ΔC/ΔE, $)	Mean effectivenessa (SE)	ICER (ΔC/ΔE, $)
MC-CBT	348a,b	5.61a.c	–	5.51a,c	–	5.50a,b	–	5.68a,b	–
	(9.32)	(0.01)		(0.02)		(0.01)		(0.01)
EDU	457b,c	5.33b.c	Economically dominated	5.31b,d	Economically dominated	5.20b,c	Economically dominated	5.25b,c	Economically dominated
	(10.37)	(0.01)		(0.02)		(0.005)		(0.01)
S-CBT	644a,c (19.20)	5.49a.c (0.01)	Economically dominated	5.35a,d (0.02)	Economically dominated	5.42a,c (0.01)	Economically dominated	5.36a,c (0.01)	Economically dominated

Note. Standard errors of the averages are presented in parentheses.

^aStatistically significant difference between MC-CBT and S-CBT averages at the .001 level.

^bStatistically significant difference between MC-CBT and EDU averages at the .001 level.

^cStatistically significant difference between S-CBT and EDU averages at the .001 level.

^dStatistically significant difference between S-CBT and EDU averages at the .05 level.

Cost-Effectiveness Analysis Results for Percentage of Patient Responders

As shown in Table 5, we found that, on average, the MC-CBT condition achieved a higher percentage of patient responders relative to both the S-CBT and EDU conditions. Sixty-one percent of patients in the MC-CBT condition were classified as responders at immediate post-treatment based on patient-reported CGI scores. However, only 55% of patients in the S-CBT condition and 44% of patients in the EDU condition were classified as responders (based on patient-reported CGI scores immediate posttreatment). Comparable results were found when scores were based on physician-reported CGI scores, and these results held at 6 months posttreatment with the MC-CBT outperforming both the S-CBT and EDU conditions at a lower per-patient treatment cost.

Table 5.

Cost-effectiveness analysis for percentage of responders on CGI, immediate and 6 months

Condition	Mean total cost per patient ($, 2016) (SE)	Immediate intent-to-treat (ITT) outcomes				6-Month intent-to-treat (ITT) outcomes
		Percentage of responders (based on patient-reported CGI)		Percentage of responders (based on physician-reported CGI)		Percentage of responders (based on patient-reported CGI)		Percentage of responders (based on physician-reported CGI)
		Mean effectivenessa (SE)	ICER (ΔC/ΔE, $)	Mean effectivenessa (SE)	ICER (ΔC/ΔE, $)	Mean effectivenessa (SE)	ICER (ΔC/ΔE, $)	Mean effectivenessa (SE)	ICER (ΔC/ΔE, $)
MC-CBT	348a,b	61.0%a,b	–	55.7%a,b	–	56.6%a.b	–	58.4%a,b	–
	(9.32)	(0.01)		(0.004)		(0.01)		(0.004)
EDU	457b,c	43.5%b,c	Economically dominated	40.4%b,c	Economically dominated	47.0%b	Economically dominated	44.8%b,c	Economically dominated
	(10.37)	(0.005)		(0.004)		(0.01)		(0.005)
S-CBT	644a,c (19.20)	54.5%a,c (0.01)	Economically dominated	50.6%a,c (0.004)	Economically dominated	48.1%a (0.01)	Economically dominated	51.6%a,c (0.005)	Economically dominated

Note. Standard errors of the averages are presented in parentheses.

^aStatistically significant difference between MC-CBT and S-CBT averages at the .001 level.

^bStatistically significant difference between MC-CBT and EDU averages at the .001 level.

^cStatistically significant difference between S-CBT and EDU averages at the .001 level.

Cost-Effectiveness Acceptability Curve (CEAC) Results

Although the MC-CBT condition economically dominated the alternative conditions across all outcomes based on the means, we still calculated CEACs via a nonparametric bootstrap method to account for the variability in the cost and effectiveness estimates and determine the probability that MC-CBT would be optimal. The findings from this analysis were similar for immediate post-treatment and 6-months posttreatment outcomes, and we present the results from the 6-month posttreatment analysis here. Specifically, Fig. 1 presents the CEACs for the CGI score outcome at 6-month posttreatment and Fig. 2 presents the CEACs for the percentage of patient responders at 6 months posttreatment. Based on the CEACs, the probability that MC-CBT is the cost-effective option for the 6-month posttreatment CGI score and percentage of patient responders is close to 100% at WTP values less than approximately $400 and $700, respectively. For higher WTP values, the probability that MC-CBT is the cost-effective option decreases as the EDU condition become a viable alternative. It should be noted that across all WTP values, the S-CBT condition is not a cost-effective option relative to MC-CBT.

Fig. 1.

Patient-reported CGI Score, 6 months posttreatment.

Fig. 2.

Percentage of patient responders, based on patient-reported CGI score 6 months posttreatment.

The plot of the incremental cost and effectiveness differentials for each outcome are shown in Figs. 3 and 4. These graphs further illustrate the findings of the CEAC analysis in that the joint incremental cost and effectiveness differentials of EDU and S-CBT relative to MC-CBT reveal both higher cost with lower effectiveness.

Fig. 3.

Cost and effectiveness differentials for patient-reported CGI scores at 6 months posttreatment.

Fig. 4.

Cost and effectiveness differentials for percentage of patient responders, based on patient-reported CGI score 6 months posttreatment.

Discussion

This study compares the economic value a novel low intensity, home-based version of CBT for patients suffering from refractory IBS, with a standard version delivered one-to-one in an outpatient setting relative to the nonspecific comparator. Our motivation for developing MC-CBT was to reduce costs—both personal and economic—and increase the efficiency of treatment delivery through a more efficient format for patients and other stakeholders (payers, providers, employers) who share the enormous burden of IBS. If MC-CBT can achieve similar outcomes compared to S-CBT at a lower per-patient cost, then MC-CBT may be a more attractive option for providers and patients who seek effective low-cost treatment options for managing notoriously intractable GI symptoms with conventional medical treatment that are largely regarded as unsatisfactory for both patient and provider [24].

The cost analysis reveals that patients in the MC-CBT condition received, on average, less direct care from their provider (measured in direct treatment hours) than patients in the S-CBT condition. This was primarily due to the design of the MC-CBT treatment protocol that required assigned patients to receive fewer CBT individual sessions than the S-CBT condition. On average, patients in the MC-CBT condition received the equivalent of 5.6 total treatment sessions (including 1.9 telephonic sessions) for a total of 6.9 h of therapy. In comparison, patients in the S-CBT condition received 8.8 total treatment sessions for a total of 12.3 h of therapy. The cost analysis revealed that MC-CBT, with its combination of individual CBT and brief telephone checkups, achieved cost savings compared to S-CBT.

Given that MC-CBT was less costly on a per-patient basis due to its low intensity, the question remains as to whether MC-CBT can achieve desired results on the selected posttreatment outcome measure of global symptom improvement—that is, is MC-CBT a cost-effective treatment option relative to the standard S-CBT approach. The results of the cost-effectiveness analysis showed that at immediate posttreatment MC-CBT economically dominated both S-CBT and EDU conditions by producing better average patient outcomes on CGI scores and percentage of patient responders at a lower cost. Importantly, these findings were sustained at 6 months post-treatment with MC-CBT continuing to outperform both the S-CBT and EDU conditions on all selected outcomes. These findings indicated that MC-CBT is a cost-effective option for providers allowing delivery of effective care at a lower cost relative to both behavioral and pharmacological alternatives. The intervention costs of CBT delivered via the web with 2.5 h of therapist support- or therapist-delivered telephone CBT (8 h of support) as part of the ACTIB trial [25] was $316.75 and $1,013, respectively (£98 per session [26]), although neither condition demonstrated the immediate and sustained effects (global improvement of IBS symptoms, CGI; IBS symptom severity reduction, IBS-SSS) as either CBT conditions featured in the IBSOS [27]. With respect to pharmacological alternatives with FDA claims for IBS, the current Wholesale Acquisition Cost for a 30-day supply currently ranges from $312–438 (lubiprostone, linaclotide) to $1195 (eluxadoline) to $1250 (alosetron). A 14-day supply of rifaximin is $11,950. Beyond affordability, the brevity (i.e., four sessions) of MC-CBT achieves enduring relief of cardinal GI symptoms with minimal erosion that does not characterize pharmacological treatments [28, 29] (and traditional CBT pain regimens [30]) once treatment ends. The cost-effectiveness data presented here, coupled with the durability of treatment effects [27], makes MC-CBT one of the most powerful and practical behavioral pain treatments most of which have a modest impact on somatic symptoms for which patients seek care [31].

Our study had notable limitations. First, our cost analysis relied on the judgement of the study team as to which activities were primarily research-related and which would be used in delivering CBT in routine clinical practice. To minimize this issue, the study team included clinical researchers who provided expertise about how such interventions would be implemented in research and clinical settings. We are confident in their judgment as to all researchers we relied on for data maintained a part-time clinical practice that provided a sound basis for gauging the research versus clinical costs. Except for the condition-specific session formats (face-to-face versus telephonic), each treatment condition incurred the same unit costs for specific treatment activities; therefore, any miscalculations in this task will have no differential effect across treatment conditions and should not affect the cost-effectiveness analysis. Second, although we have attempted to identify activities and costs that would be part of real-world treatment practice from the provider perspective, in reality this study was part of a randomized controlled trial and followed the study’s rigorous protocol for treatment provision. We expect that patients’ treatment adherence may be better in a study trial compared to real-world practice where retention strategies are less uniformly implemented, and, therefore, patients may have been seen more frequently during their treatment episode in the study; therefore, our cost estimates may be an upper bound of the actual treatment costs providers could incur for delivering these services. On the other hand, we should note that less treatment adherence may also lead to poorer treatment outcomes than we observed in this trial; however, we do not expect that the impact on treatment adherence would affect treatment conditions differently; therefore, the differential effect across treatment conditions should not affect the cost-effectiveness analysis. In addition, we only included the costs of delivering treatment interventions under study. We did not take a broader perspective and include additional health care and other non-therapy costs that may have been incurred. It is possible that receipt of the studied interventions could affect participants’ use of additional services resulting in different cost findings from this broader perspective. Finally, our cost-effectiveness analysis was limited to the three treatment conditions included in this study—MC-CBT, S-CBT, and EDU. The inclusion of additional interventions or an alternative set of strategies would provide different comparisons between cost and effectiveness, and different cost-effectiveness results. Furthermore, our cost-effectiveness analysis was limited to four selected outcomes which served as the a priori primary outcome of the study. The results of this analysis may differ if other clinical outcomes were considered. Our choice of endpoint (global relief of IBS symptoms based on the IBS version of the CGI) was based on expert opinion for clinical trials for both IBS [20] and chronic pain [32]. We recognize that the establishment of an acceptable endpoint to multiple stakeholders (patients, providers, policymakers, payers) for a multisymptom disorder like IBS with great heterogeneity and impacts (e.g., quality of life, wellbeing) across multiple domains is a work in progress [33], but we are confident that our results reflect best clinical practice using recommended outcome measures featured in other behavioral and drug trials. While the CGI has its limitations (single-item measure of global symptom improvement), its focus on the perceived improvement of GI symptoms captures the clinical meaningfulness of treatment benefit from the patient’s perspective than endpoints that infer such benefit from a numerical change in symptom severity. Further research is needed to characterize the clinical significance of therapeutic benefit derived from patient-reported outcomes [19, 34, 35]. While unsettled, we do not believe this issue detracts from the overall significance of our findings. In our view, it is likely that the societal valuation for the better outcome—for example, an approximate 10 percentage point differential in patient responders (as derived by the CGI score)—within the MC-CBT group exceeds the estimated $348 per-patient cost.

Despite these limitations, the present study makes an important contribution to the literature because it provides an economic analysis of an innovative, low-intensity approach for delivering CBT for IBS. Our study found MC-CBT performed better at immediate (2 weeks after treatment ended) and 6 months posttreatment on the selected outcomes and cost-effectiveness than S-CBT. At a general level, these data extend the parameters of the therapeutic value of minimal contact CBT for IBS featured in the IBSOS. Previous research has shown that it is not only efficacious [12] for more severe IBS but demonstrates durability, lack of decay [36], rapidity of action [37], benign side effect profile, which are unmatched by other psychological (e.g., traditional CBT delivered in an office setting face to face, telephonically or web-based; hypnosis), medical, and dietary IBS treatments. Overall, these results strongly demonstrate the economic and clinical value of MC-CBT, making it a compelling choice for providers when deciding on treatments for IBS patients with moderate to severe symptoms. For health care payers, data provide much-needed guidance on a novel therapy that strikes an unique balance between efficacy (e.g., improvement of refractory symptoms) and efficiency (few office visits, self-directed learning, convenience) deemed critical in a competitive value-based health care environment particularly in the COVID era when extended outpatient, one-to-one encounters are limited for reasons beyond clinical efficacy and digital therapeutics have not achieved an efficacy profile commensurate with their scalability potential and convenience [38, 39].

Compliance with Ethical Standards

Authors’ Statement of Conflict of Interest and Adherence to Ethical Standards The authors declare that they have no conflict of interests.

Author Contributions The authors had full access to all of the data in the study and take responsibility for the integrity of the data and the accuracy of the data analysis. Study concept and design: Lackner, Jaccard, Dunlap; acquisition, analysis, or interpretation of data: All authors. Drafting of the manuscript: Dunlap and Lackner. Critical revision of the manuscript for important intellectual content: All Authors. Obtained funding: Lackner. Administrative, technical, or material support: All authors. Study supervision: All Authors.

Ethical Approval All procedures performed in this study were in accordance with the ethical standards of the institutional research committee and with the 1964 Helsinki declaration and its later amendments or comparable ethical standards.