A Health Economic Evaluation of Digital Digestive Care Management

ABSTRACT

Objectives: Chronic gastrointestinal disorders are common and costly for employers. We sought to evaluate the effects of an employer-sponsored digital digestive care program on health care spending.

Study Design: Retrospective controlled cohort study.

Methods: Using propensity score matching of participants and nonparticipants, we evaluated the health care spending of users of a digital digestive care program vs matched controls.

Results: At baseline, the mean (SD) age of the 347 participants and 1041 matched controls was 44 (10) years, 87% were female, and total mean (SD) annual health care spending was $8884 ($12,884) per member per year (PMPY). The mean program cost was $345 PMPY. Our results show savings of $5.87 for every dollar invested ($2026 savings PMPY / $345 program cost PMPY), for a net return on investment of $4.87 for every $1.00 invested after subtracting program costs.

Conclusions: Digital digestive care is promising as a cost-saving employer-sponsored benefit.

Am J Manag Care. 2026;32(1):In Press

Takeaway Points

Chronic gastrointestinal disorders are common and costly for employers. This study evaluated the effects of virtual digestive care management, which includes symptom tracking, personalized medical nutrition therapy, health coaching, and targeted education. In the year following the implementation of this program, annual health care expenditures for participants were $2026 lower compared with nonparticipants—an 18% savings. Practical takeaways include the following:

• We discuss methods to design a credible evaluation of return on investment in an employer-sponsored benefit program focused on gastrointestinal health.
• We identify factors important to the success of a digital digestive health intervention in an employed population demonstrating improved care and control of health care costs.
• When considering increasing access to specialty gastrointestinal care, virtual digestive care management may offer a cost-saving alternative or supplement to traditional care.

Chronic gastrointestinal (GI) conditions affect approximately 1 in 3 US adults and result in more than $135 billion (2015 US$) in health care expenditures annually.¹ Employers pay for both the direct medical costs and the significant productivity losses associated with these conditions given that chronic GI symptoms peak in midlife, affecting working adults.^2-5 Employers, therefore, are seeking effective, evidence-based interventions to support their employees and dependents with chronic GI symptoms.

Recent independent evaluations of employer-sponsored health interventions have found conflicting evidence of their clinical or economic benefits, with many programs actually driving up health care spending.^6-9 It is notable that among employer-sponsored programs with demonstrated clinical and economic benefit (eg, early access to mental health services,⁹ virtual physical therapy⁷), the target populations are symptomatic, in contrast to those interventions serving often asymptomatic populations such as those with diabetes.⁶ Chronic GI conditions such as irritable bowel syndrome and gastroesophageal reflux disease can be highly symptomatic—prompting patients to seek care and engage in symptom-focused interventions.¹⁰ Evidence on the effectiveness of care management programs for digestive disorders is emerging but remains limited.¹¹ Prior studies have shown that adults with GI symptoms (across genders, ethnic groups, and social vulnerability strata) are willing to use an employer-sponsored digital digestive care management program.^12,13 However, there have been no comprehensive assessments of the economic outcomes of employer-sponsored programs directed at workers with digestive disorders.

The purpose of this study was to evaluate the effects of an employer-sponsored digital digestive care program—which has been previously shown to improve the symptoms of adults with chronic GI symptoms,^12,13 support health equity,¹⁴ and decrease sick day usage¹⁵—on participants’ total medical and pharmacy claims expenditures after 1 year of exposure to the intervention.

METHODS

Intervention

The digital digestive health program evaluated in this study has 4 key components: symptom tracking, personalized medical nutrition therapy, health coaching, and targeted education. It has been described in detail elsewhere.^12,13 Briefly, a smartphone app enables users to monitor their symptoms over time. Licensed registered dieticians provide evidence-based, one-on-one medical nutrition therapy, including education on the management of common GI conditions. Certified health coaches provide one-on-one support with goal setting, self-management skill building, and use of GI disease management resources. The program provides participants with personalized care plans that include targeted education on their symptoms, conditions, and relevant lifestyle interventions. Users of this program have had significant reductions in symptoms and symptom severity.^12,13 Bravata et al found that at baseline, 79.5% of users reported at least moderate GI symptom severity for at least 1 symptom while at the end of the intervention, only 47.8% of participants reported moderate or severe symptoms and 16.0% of participants reported no symptoms.¹³

Participants

The digital digestive care program was offered to 10,974 employees and adult dependents of a large public school system in the Southeast US as an employee health benefit. Employees were made aware of the program via organizational emails, direct mailers, and internal promotions (eg, flyers and webinars).

Program Launch and Evaluation Period

The digestive care program was launched on February 1, 2023. We evaluated changes in total medical and pharmacy paid amounts between the baseline period (12 months before the program launch date or February 1, 2022, through January 31, 2023) and the intervention period (12 months after the program launch date or February 1, 2023, through January 31, 2024).

Inclusion and Exclusion Criteria

Participants (n = 372) were initially considered for inclusion in the study if they registered for the program within the first 3 months after its launch (February 1, 2023, to April 30, 2023) and had continuous medical coverage for the entire baseline and intervention periods. Nonparticipants (n = 6893) were included if they had continuous medical coverage for the entire baseline and intervention periods but did not register for the program.

Participants and nonparticipants were excluded if they were younger than 18 years or older than 64 years at any time during the study, required hospice or skilled nursing facility care, or had greater than the 99th percentile in pharmacy costs (ie, $55,000), medical costs (ie, $100,000), or professional costs (ie, $25,000) in either the preprogram or postprogram periods. Nominal dollar amounts were used for each year’s analysis.

Matching

We propensity score matched participants and nonparticipants within the same public school system based on age, self-reported gender, preprogram total health care costs broken out by claim category (ie, inpatient, emergency department [ED], other hospital, and outpatient nonfacility and pharmacy), preprogram GI-specific medical costs (to help address differences in GI conditions with different care needs such as functional conditions and cancer), postprogram HHS Hierarchical Condition Category (HHS-HCC) scores (the risk adjustment model used by HHS for the individual and small group markets in response to the Affordable Care Act, which matches on the basis of age, sex, acute and chronic conditions, and combinations of comorbid conditions based on the preceding year’s diagnosis codes),¹⁶ and a postprogram GI condition flag (whether the person had a diagnosed GI condition or not).

The matching approach was N:1 matching without replacement, meaning multiple nonparticipants could be matched to the same participant and each nonparticipant could only be matched to 1 participant. The matching frequency weight was 1 for every treated unit (participant), and 1/n_i for each of the matched nonparticipants, where n_i was the number of matched nonparticipant(s) for participant i.

Data Collection

We collected demographic data from employer eligibility files and participant engagement from the app. Medical and pharmacy claims were provided by the employer’s payer; these included the top 2 diagnostic fields, from which we determined GI and other conditions. We computed HHS-HCC scores and determined clinical conditions from claims. The HHS-HCC model was chosen because the study population was similar to the commercial population used in the HHS-HCC model (as opposed to the Medicare population, where risk is adjusted using the CMS-HCC model). We deployed SAS code to simulate the HHS-HCC model as provided by HHS.¹⁶

Analysis

We utilized a retrospective cohort study design deploying 3:1 propensity score matching and difference-in-differences (DID) methodology to quantify the change in total health care costs from the baseline period to the intervention period for participants vs nonparticipants. We used a generalized linear model to test the significance of the DID.

After propensity score matching, we used a DID method to estimate the savings (net cost change) for the participants. The variables a, b, c, and d in Table 1 represent the weighted mean of the outcome variable (medical and pharmacy per-member per-year [PMPY] costs where PMPY = 12 × [total cost / member months]) in each cell, weighted by a matching frequency weight. The savings were estimated through DID by the equation: Savings = (d – c) – (b – a). Changes in utilization of services (inpatient care, emergency care, other hospital-based care, and outpatient care) were calculated in a similar manner.

Return on investment (ROI) calculation. Intervention program fees were paid by the employer. ROI was calculated via the equation: ROI = (savings PMPY) / (mean program fee PMPY). The mean program fee PMPY was calculated via the equation: (total program fees in the intervention period) / (number of unique participants billed for the program in the intervention period). For the program to be net ROI positive with this ROI calculation method,¹⁷ the ROI would need to be greater than 1.

Human subjects approval. Given that all data were routinely collected as part of the condition management program, this protocol was considered exempt by the institutional review board (Vanderbilt #241271, April 2023). The analysis and presentation of data conform to the Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) guidelines (STROBE checklist in eAppendix A [eAppendices available at ajmc.com]).

RESULTS

Matching and Demographics

After matching and exclusion, the study population included 347 participants and 1041 propensity score–matched nonparticipants. Table 2 compares the characteristics of participants and nonparticipants at baseline before exclusion criteria were applied and before propensity score matching. All matching covariates were balanced with an absolute value standardized difference less than or equal to 0.1. At baseline, the mean (SD) age of participants was 44 (10) years, 87% were female, and their mean annual total health care spending was $8884 per person. Before matching, the participants were significantly more likely to be female, have a mental health condition, and have any GI diagnosis than nonparticipant controls (Table 2). After matching, participants and controls were similar across all these characteristics. This suggests that the propensity scoring adequately matched participants and nonparticipants.

Program Utilization

Participants used the app for a variety of activities (eg, symptom logging, reviewing educational materials, interacting with their care team). Between 0 and 90 days post registration, the participant population had a mean (SD) of 14.6 (48.0) total interactions with the program. Nonparticipants had no interactions with the app or program staff.

Cost Savings

Baseline mean (SD) total paid health care expenditures PMPY were comparable for participants and nonparticipants ($8884 [$12,884] vs $8757 [$7080]; P = .85), respectively (Table 1 and Figure 1). During the intervention period, the mean (SD) total paid health care costs PMPY were $9442 ($12,047) among participants and $11,341 ($7607) among nonparticipants.

Overall, although total health care costs increased for both participants and nonparticipants from the baseline period to the intervention period (Table 3), participants experienced a significantly lower net increase of $2026 PMPY (ie, [$11,341 – $8757] – [$9442 – $8884]) compared with nonparticipants (–18%; P = .004) (Figure 1). The bulk of the cost reductions ($2534 PMPY) came from reductions in facility claims—inpatient ($932 PMPY), ED ($594 PMPY), and other hospital ($1008 PMPY)—offset by moderately higher outpatient nonfacility costs ($430 PMPY) and slightly higher pharmacy costs ($78 PMPY) among participants compared with nonparticipants (Figure 2 and eAppendix B).

In sensitivity analyses, we evaluated using utilization as a matching covariate, varying outlier thresholds, and excluding individuals with pregnancy claims. Results were similar to the main reported findings (eAppendix C).

Health Care Utilization

Compared with nonparticipants, participants had decreases of 339 and 527 per 1000 in inpatient and ED utilization, respectively (P < .01 for all). This corresponded to decreases of 25% and 27% vs nonparticipant comparisons for inpatient and ED utilization, respectively. Additionally, nonparticipants had more ED visits for GI issues (eg, nonspecific abdominal pain, nausea and vomiting) in the postintervention period than participants.

ROI

The mean program cost PMPY was $345. The program achieved an ROI of $5.87 saved for every dollar invested ($2026 savings PMPY / $345 program cost PMPY), with a net savings of $4.87 for every dollar invested after subtracting program costs from the numerator of the ROI metric.

DISCUSSION

This study demonstrates that an employer-sponsored digital digestive management program can have a significant impact on total medical and pharmacy utilization and associated health care spending. The study had 3 primary results. First, we found that in the year following program implementation, health care expenditures for participants were $2026 lower PMPY compared with nonparticipants—an 18% savings compared with controls. Thus, the program achieved an ROI of 5.87:1, with a net savings of $4.87 for every dollar invested after subtracting program costs.

When considering the results of this analysis in the context of recent evaluations of employer-sponsored health interventions with demonstrated savings (such as those for patients with mental health concerns⁹ and musculoskeletal issues⁷), we hypothesize that the savings associated with the digestive disease management program may have in part resulted from its ability to reduce worrisome digestive symptoms. Previously, the intervention has been associated with significant reductions in GI symptoms and disease severity^12,13 and decreases in sick days and workplace absenteeism.¹⁵ This is not surprising because the intervention includes symptom tracking and the care team works to provide users with personalized nutritional plans aimed at improving their symptoms. The promising findings from this analysis warrant an assessment of savings stratified by types of symptom improvement.

We note that the baseline medical spending of participants was relatively high for a commercially insured population at $8884 PMPY. By comparison, a 2021 study of commercially insured individuals found mean annual spending to be $4774.¹⁸ It may be that the savings and associated ROI demonstrated in this evaluation was related to the relatively high baseline health care spending of the participants. This warrants further evaluation of the intervention in populations with lower baseline health services utilization.

Second, the observed savings were largely driven by decreased utilization and spending for inpatient admissions and emergency care. This decrease in health care utilization and associated savings was observed much earlier (in the first year after implementation of the intervention) compared with other employer-sponsored interventions.⁸ This may have been a result of engaging participants earlier in the care journey and preventing clinical deterioration while shifting care to less expensive outpatient venues.

Third, participants had slightly higher pharmacy costs than nonparticipants ($78 PMPY). A recent evaluation of employer-sponsored interventions found that several programs associated with reduced health care spending demonstrated improvements in medication adherence for employees.¹⁹ Patients with chronic GI conditions may be prescribed a variety of low-cost therapies for symptom control (eg, antacids, fiber supplements) and high-cost medications that directly control their underlying disease processes (eg, biologics for inflammatory bowel disease). The current analysis did not provide a detailed assessment of both OTC and prescription drug use, but future evaluations should specifically address the extent to which the intervention impacts medication adherence broadly.

Limitations

This study had 3 key limitations. First, it evaluated the implementation of a digestive care management program in a single employer in a single state at 1 year. Future evaluations should include a wide range of commercially insured populations across the US and evaluate savings beyond 1 year. Second, it evaluated predominantly female employees, thus limiting the generalizability of its findings. Two factors may contribute to this disparity: Many GI conditions characterized by chronic digestive symptoms, such as irritable bowel syndrome, are more common in women than in men,¹⁰ and women may engage more in digital health interventions than men.^12,13 Third, although propensity score matching enables the ability to balance groups based on observable characteristics, it does not account for unobserved or unmeasured cofounders that could drive participation with a care management program. Given that employers are typically reluctant to sponsor randomized controlled trials of their benefit programs, a future analysis might evaluate similar employees with access to the program compared with those without such access.²⁰

CONCLUSIONS

Overall, we found that participants of a digital digestive health intervention had 18% lower costs compared with nonparticipants 1 year after program implementation while achieving a significantly positive ROI. Such a program appears promising for improving care related to digestive health while helping to control rising health care costs and warrants expanded evaluation across other employers and other populations.

Data Availability

Data are available from the authors upon reasonable request.

Author Affiliations: Vanderbilt University Medical Center (MS), Nashville, TN; Metro Nashville Public Schools (MS, DH), Nashville, TN; Cylinder Health (PCH, HL, DB), Chicago, IL; Benegration (JH-S), Elkins Park, PA; NYU Grossman School of Medicine (HL), New York, NY; Stanford University (DB), Stanford, CA; Brown University School of Public Health (CMW), Providence, RI; Johns Hopkins Bloomberg School of Public Health (KDF, EFD, RZG), Baltimore, MD; Johns Hopkins Carey Business School (KDF), Baltimore, MD.

Source of Funding: This project was supported by the Metro Nashville Public Schools and Cylinder Health.

Author Disclosures: Mr Ho and Dr Liu are employed by and own stock in Cylinder Health, which provided the solution evaluated in this article. Dr Liu also has a patent pending regarding Cylinder Health and presented this work at the Business Group on Health Annual Conference. Mr Harris-Shapiro is a consultant to Vanderbilt University Medical Center and Metro Nashville Public Schools. Dr Bravata is a consultant to Cylinder Health, received payment from Cylinder Health for her involvement in the preparation of this article, and owns stock in Cylinder Health. Drs Frick, Drabo, and Goetzel are employed by Johns Hopkins University, which received payment from Cylinder Health for support of the study design, data analysis, and manuscript preparation. Dr Drabo also participated in remunerated consultation activities with Pfizer Inc outside of this work. The remaining authors report no relationship or financial interest with any entity that would pose a conflict of interest with the subject matter of this article.

Authorship Information: Concept and design (MS, PCH, JH-S, HL, DB, CMW, KDF); acquisition of data (MS, PCH, DH, JH-S); analysis and interpretation of data (PCH, JH-S, HL, DB, CMW, KDF, EFD, RZG); drafting of the manuscript (PCH, HL, EFD, RZG); critical revision of the manuscript for important intellectual content (MS, DH, HL, DB, CMW, KDF, EFD, RZG); statistical analysis (PCH, CMW, EFD, RZG); provision of patients or study materials (MS, DH); obtaining funding (HL); administrative, technical, or logistic support (JH-S, DB); and supervision (HL).

Address Correspondence to: Hau Liu, MD, Cylinder Health, 2045 W Grand Ave, Ste B, PMB 37767, Chicago, IL 60612. Email: hliu@cylinderhealth.com.

REFERENCES

1. Peery AF, Crockett SD, Murphy CC, et al. Burden and cost of gastrointestinal, liver, and pancreatic diseases in the United States: update 2018. Gastroenterology. 2019;156(1):254-272.e11. doi:10.1053/j.gastro.2018.08.063

2. Kuenzig E, Lebenbaum M, Mason J, et al. Costs of missed work among employed people with inflammatory bowel disease: a cross-sectional population-representative study. Int J Popul Data Sci. 2022;7(3). doi:10.23889/ijpds.v7i3.1918

3. Paré P, Gray J, Lam S, et al. Health-related quality of life, work productivity, and health care resource utilization of subjects with irritable bowel syndrome: baseline results from LOGIC (Longitudinal Outcomes Study of Gastrointestinal Symptoms in Canada), a naturalistic study. Clin Ther. 2006;28(10):1726-1735; discussion 1710-1711. doi:10.1016/j.clinthera.2006.10.010

4. Bytzer P, Langkilde LK, Christensen E, Meineche-Schmidt V. Work productivity improvement after acid suppression in patients with uninvestigated dyspepsia. Dan Med J. 2012;59(7):A4461.

5. Dean BB, Aguilar D, Barghout V, et al. Impairment in work productivity and health-related quality of life in patients with IBS. Am J Manag Care. 2005;11(suppl 1):S17-S26.

6. Digital Diabetes Management Solutions. Peterson Health Technology Institute; March 2024. Updated April 19, 2024. Accessed January 5, 2025. https://phti.org/assessment/digital-diabetes-management-tools/

7. Virtual Musculoskeletal (MSK) Solutions. Peterson Health Technology Institute; June 2024. Updated October 10, 2024. Accessed January 5, 2025. https://phti.org/assessment/virtual-msk-solutions/

8. Mattke S, Liu HH, Caloyeras JP, et al. Do workplace wellness programs save employers money? RAND. January 9, 2014. Accessed May 21, 2024. https://www.rand.org/pubs/research_briefs/RB9744.html

9. Claims cost impact of behavioral health services: 2021 cost efficiency measurement. Lyra Health. July 2024. Accessed January 10, 2025. https://www.lyrahealth.com/resources/white-paper/lyra-delivers-the-highest-roi-in-mental-health-care/

10. Chey WD, Kurlander J, Eswaran S. Irritable bowel syndrome: a clinical review. JAMA. 2015;313(9):949-958. doi:10.1001/jama.2015.0954

11. Zia JK, Barney P, Cain KC, Jarrett ME, Heitkemper MM. A comprehensive self-management irritable bowel syndrome program produces sustainable changes in behavior after 1 year. Clin Gastroenterol Hepatol. 2016;14(2):212-219.e1-2. doi:10.1016/j.cgh.2015.09.027

12. Bravata D, Liu H, Colosimo MM, Bullock AC, Commons E, Pimentel M. Digital disease management programme reduces chronic gastrointestinal symptoms among racially and socially vulnerable populations. BMJ Open Gastroenterol. 2024;11(1):e001463. doi:10.1136/bmjgast-2024-001463

13. Bravata DM, Liu H, Colosimo MM, Bullock AC, Commons E, Pimentel M. Improvements in digestive symptoms after participation in an app-based chronic digestive disease management program: a prospective cohort evaluation. Cureus. 2024;16(8):e66941. doi:10.7759/cureus.66941

14. Varma S, Black A, Commons E, Ho PC, Bravata DM, Liu H. The impact of a digital digestive health program and telehealth visits in socially vulnerable populations: cohort evaluation. JMIR Form Res. 2025;9:e70748. doi:10.2196/70748

15. Shepherd M, Ho PC, Harris-Shapiro J, et al. The effects of a digital digestive care management program on employee absenteeism: a case-control study. J Occup Environ Med. 2025;67(9):e612-e615. doi:10.1097/JOM.0000000000003441

16. HHS-developed risk adjustment model algorithm “do it yourself (DIY)” software instructions for the 2023 benefit year: August 22, 2023 update. CMS. Accessed December 11, 2024. https://www.cms.gov/files/document/cy2023-diy-instructions-08222023.pdf

17. Return on investment estimation. Agency for Healthcare Research and Quality. Accessed December 11, 2024. https://www.ahrq.gov/sites/default/files/wysiwyg/professionals/systems/hospital/qitoolkit/combined/f1_combo_returnoninvestment.pdf

18. Duffy EL, Green S, Randall S, Trish E. Beyond average spending: distributional and seasonal commercial insurance trends, 2012-2021. Am J Manag Care. 2024;30(9):415-420. doi:10.37765/ajmc.2024.89600

19. Digital Hypertension Management Solutions. Peterson Health Technology Institute; October 2024. Updated November 26, 2024. Accessed January 10, 2025. https://phti.org/assessment/digital-hypertension-management-solutions

20. Mattke S, Liu H, Caloyeras JP, et al. Workplace Wellness Programs Study: Final Report. RAND Health; 2013. Accessed May 25, 2025. https://www.rand.org/content/dam/rand/pubs/research_reports/RR200/RR254/RAND_RR254.pdf