ABSTRACT
Objective
Although concerns have been raised that glucagon‐like peptide‐1 receptor agonists (GLP‐1RAs) may increase the risk of bowel obstruction or ileus, evidence remains inconclusive. In this context of uncertainty, we examined the frequency and determinants of GLP‐1RA discontinuation after hospitalization for bowel obstruction/ileus.
Methods
We conducted a retrospective cohort study using a nationwide Japanese health insurance claims database (DeSC). We identified individuals aged ≥18 years with type 2 diabetes who were hospitalized for bowel obstruction/ileus between 2016 and 2023 while receiving GLP‐1RA therapy. The primary outcome was GLP‐1RA discontinuation, defined as no prescription at discharge or within 6 months of discharge. Factors potentially associated with discontinuation were grouped into (1) general factors, (2) severity of bowel obstruction/ileus, and (3) potential therapeutic benefit of GLP‐1RA therapy. Risk differences (RDs) for discontinuation were estimated using multivariable linear regression with robust standard errors.
Results
GLP‐1RAs were discontinued in 86 (43.9%) of the 196 patients analyzed. Multivariable analysis revealed that surgery for bowel obstruction during hospitalization (RD = 32.1%; 95% confidence interval [CI] 11.2% to 53.0%, p = 0.003), longer length of stay (≥15 days vs. ≤7 days; RD = 32.4%; 95% CI 11.2% to 53.5%, p = 0.003), and heart failure (RD = 17.6%; 95% CI 2.1% to 33.2%, p = 0.027) were associated with GLP‐1RA discontinuation.
Conclusions
Nearly half of the patients discontinued GLP‐1RAs after a bowel obstruction/ileus event. While GLP‐1RAs were more likely to be discontinued after severe events, their potential therapeutic benefits appeared less frequently considered in continuation decisions.
Keywords: drug‐related side effects and adverse reactions, glucagon‐like peptide‐1 receptor agonists, ileus, intestinal obstruction
1. Introduction
Glucagon‐like peptide‐1 receptor agonists (GLP‐1RAs) are widely used in the management of type 2 diabetes owing to their beneficial effects on cardiovascular outcomes, all‐cause mortality, glycemic control, and weight loss [1, 2]. Since their early clinical use, concerns have been expressed regarding a potentially elevated risk of bowel obstruction or ileus (i.e., impaired propulsive bowel activity in the absence of mechanical obstruction) associated with GLP‐1RA use. In 2012, the Pharmaceuticals and Medical Devices Agency (PMDA) of Japan reported three cases of bowel obstruction or ileus potentially associated with liraglutide use [3]. Similar safety signals were subsequently reported by the European Medicines Agency [4] and the US Food and Drug Administration [5]. In Japan, bowel obstruction has been listed as a serious adverse drug reaction in the product information for several GLP‐1RA formulations since 2012, including liraglutide [6], exenatide [7], and dulaglutide [8], and the same labeling was added to semaglutide, tirzepatide, and lixisenatide in 2025. However, although one observational study reported an increased risk of bowel obstruction/ileus among GLP‐1RA users [9], more recent observational studies have reported no significant differences [10, 11, 12], and a recent meta‐analysis of randomized controlled trials found no significant association between GLP‐1RA use and bowel obstruction or ileus [13]. These results suggest that although authorities have voiced concerns, GLP‐1RAs may not truly increase the risk of bowel obstruction or ileus.
Discontinuation of the drug is a common clinical practice when it is suspected of causing an adverse event [14, 15]. Generally, discontinuation is more likely when the suspected adverse event is severe [16, 17, 18]. This approach would be generally reasonable, as patients who experience a drug‐induced adverse event are often considered to have a higher risk of recurrence of similar events upon re‐exposure [19, 20]. In fact, in previously reported cases of bowel obstruction/ileus leading to hospitalization during GLP‐1RA treatment, most patients discontinued the drug following the event [3, 21, 22]. However, given the uncertainty regarding the causal relationship between GLP‐1RAs and bowel obstruction/ileus, discontinuing therapy on the basis of event severity may not be the optimal strategy. At the same time, the potential harms of discontinuation should also be considered, such as elevated cardiovascular risk, worsening of glycemic control, or inadequate weight management [23, 24, 25]. Nevertheless, physicians in Japan may be inclined to discontinue GLP‐1RA therapy after bowel obstruction/ileus events despite ongoing indications, in response to concerns raised by the PMDA or product labeling that lists bowel obstruction/ileus as serious adverse events [6, 7, 8]. In this context, we sought to investigate how frequently GLP‐1RAs are discontinued following bowel obstruction/ileus events and to better understand the factors influencing these decisions.
Using data from a Japanese health insurance claims database, we aimed to quantify the frequency of GLP‐1RA discontinuation and identify the factors associated with it among patients with type 2 diabetes who developed bowel obstruction/ileus while receiving GLP‐1RAs. Specifically, we were interested in whether the decision to discontinue GLP‐1RAs was associated with the severity of the bowel obstruction/ileus episode, the potential therapeutic benefits of continuing GLP‐1RA therapy, or both. We focused on individuals prescribed GLP‐1RAs for type 2 diabetes and excluded those treated solely for obesity to ensure a more homogenous study population and reduce indication‐related heterogeneity.
2. Methods
2.1. Study Design and Data Source
This retrospective cohort study was conducted using data extracted from the commercially available DeSC database (DeSC Healthcare Inc., Tokyo, Japan) [26]. This database contains outpatient and inpatient claims data collected from health insurers (Health Insurance for salaried employees (Kempo), National Health Insurance (Kokuho), and Advanced Elderly Medical Service System (Kouki Koureisha Iryo Seido) for individuals aged ≥75 years) between April 2014 and August 2024, covering approximately 12.5 million individuals residing in Japan. A previous study confirmed that the population of this database is broadly representative of the general Japanese population [27]. Diagnoses are recorded using the International Classification of Diseases, 10th revision (ICD‐10) codes, and medications are coded according to the World Health Organization Anatomical Therapeutic Chemical classification system. The database also contains annual health check‐up data, comprising measurements such as height, weight, blood and urine test results, and self‐reported lifestyle questionnaires.
2.2. Eligibility Criteria
We initially identified the hospitalization records of patients admitted for bowel obstruction/ileus. These conditions were defined using a combination of ICD‐10 codes (K560, K562, K565, K566, K567, or K913) as the primary diagnosis for hospitalization, in conjunction with the performance of radiography or computed tomography scan on the day before, the day of, or the day after admission. Only hospitalizations with complete follow‐up from admission through discharge were included. The hospitalization records of patients transferred to another hospital within 3 days of the initial admission were treated as a single hospitalization.
Thereafter, we restricted the population to individuals with type 2 diabetes who were receiving GLP‐1RA therapy at the time of hospitalization. Type 2 diabetes was identified using ICD‐10 codes E11, E12, E13, or E14, excluding patients with code E10 (indicating type 1 diabetes). GLP‐1RA use was defined as receipt of a prescription within 120 days prior to hospitalization. We included only the first eligible hospitalization for bowel obstruction/ileus during GLP‐1RA therapy for each individual. Other inclusion criteria were age ≥18 years, hospitalization between April 2016 and December 2023, and continuous enrollment in the database for at least 2 years prior to hospitalization to ensure an adequate look‐back period for baseline data.
2.3. Variable and Outcome
We collected data on the baseline patient characteristics, hospital characteristics, procedures performed during hospitalization, in‐hospital outcomes, and GLP‐1RA discontinuation within 6 months after discharge.
The baseline patient characteristics included age, sex, year of hospitalization, type of health insurer, class of GLP‐1RA prescribed, time since initiation of GLP‐1RAs, and concomitant use of other antidiabetic drugs (insulin, metformin, sodium‐glucose cotransporter 2 (SGLT2) inhibitors, dipeptidyl peptidase‐4 (DPP4) inhibitors, thiazolidines, sulfonylureas, alpha‐glucosidase inhibitors (AGIs), or glinides). Comorbidities included obesity or overweight, hypertension, dyslipidemia, heart failure, coronary artery disease, stroke or transient ischemic attack, chronic kidney disease, dementia, and cancer (subclassified as gastric, colonic, hepatobiliary‐pancreatic, urinary tract, and gynecological). We also determined whether the patient had undergone abdominal surgery or had experienced bowel obstruction/ileus within 2 years prior to the index hospitalization.
The in‐hospital factors included the type of admitting hospital (university hospital or not), documentation of a specific etiology or site of obstruction (ileus, small bowel obstruction, large bowel obstruction, colonic volvulus, or none of the above), performance of bowel decompression (via nasogastric tubes or long tubes), performance of surgery for bowel obstruction, intensive care unit (ICU) admission, length of stay, and in‐hospital death.
The primary outcome was GLP‐1RA discontinuation, defined as the absence of any GLP‐1RA prescription at discharge or within 6 months after discharge.
2.4. Statistical Analysis
We described the patient characteristics, procedures performed during hospitalization, and in‐hospital outcomes. Continuous variables were reported as means and standard deviations, whereas dichotomous and categorical variables were displayed as numbers and percentages. We calculated the proportion of patients who discontinued GLP‐1RAs after hospitalization. For this and all subsequent analyses, we restricted the cohort to patients who were discharged alive and had undergone at least 6 months of follow‐up without death. To assess whether GLP‐1RAs were more likely to be discontinued than other antidiabetic drugs, we compared the discontinuation proportions of GLP‐1RAs with each other antidiabetic drug class among patients who were receiving both drugs prior to hospitalization (defined as a receipt of a prescription within 120 days before admission) using exact McNemar's test. These analyses were performed separately for each antidiabetic drug.
We assessed the factors associated with GLP‐1RA discontinuation. These factors were grouped into (1) general factors (age, sex, year of admission, type of health insurer, time since GLP‐1RA initiation, dementia, history of cancer, abdominal surgery within the past 2 years, bowel obstruction/ileus within the past 2 years, and admission to a university hospital); (2) factors related to the severity of bowel obstruction/ileus (specific etiology of bowel obstruction/ileus, need for bowel decompression or surgery for bowel obstruction, ICU admission, and length of stay); and (3) factors related to the potential benefits of GLP‐1RA therapy (intensity of diabetes treatment (≤2 types of antidiabetic drugs without insulin, ≥3 types of antidiabetic drugs without insulin, and insulin use), obesity, hypertension, dyslipidemia, heart failure, coronary artery disease, stroke or transient ischemic attack, and chronic kidney disease). Both univariable and multivariable analyses were conducted using linear regression models with robust standard errors [28] to estimate the risk differences (RDs) for discontinuation and their corresponding 95% confidence intervals (CIs).
No adjustments for multiple comparisons were made owing to the exploratory nature of this study, and a two‐tailed p‐value <0.05 was considered statistically significant. Statistical analyses were performed using STATA/MP, version 19 (StataCorp LLC, College Station, TX, USA).
2.5. Subgroup Analysis
After obtaining the main results and observing that GLP‐1RAs were more likely to be discontinued in patients who underwent surgery for bowel obstruction or whose length of stay was ≥15 days, we compared the discontinuation proportions of GLP‐1RAs and other antidiabetic drugs within these subgroups. These comparisons were performed using the approach employed in the main analysis.
2.6. Sensitivity Analysis
We conducted two sensitivity analyses to examine the robustness of our findings. First, we restricted bowel obstruction/ileus hospitalizations to patients who underwent radiography or computed tomography on the day of admission, rather than allowing a 1‐day lag. Second, we included patients without 6 months of follow‐up after discharge in the discontinuation analyses, using data up to the end of follow‐up. For example, if a patient died 3 months after discharge without resuming GLP‐1RAs, they were classified as having discontinued GLP‐1RAs.
3. Results
3.1. Baseline Characteristics
We initially identified 122,749 hospitalizations for bowel obstruction/ileus. Of these, 524 involved patients with type 2 diabetes who were receiving GLP‐1RA therapy at the time of admission. After applying the eligibility criteria, 303 patients remained for the initial analysis. For the analysis of GLP‐1RA discontinuation, we further excluded those who died during hospitalization (n = 18) and those who could not be followed for 6 months after discharge (n = 89, including 33 who died within 6 months), leaving 196 patients (Figure 1).
FIGURE 1.
Flowchart depicting selection of the study participants. GLP‐1RA, glucagon‐like peptide‐1 receptor agonist.
The patients’ mean age was 78 years (standard deviation, 9.9), and 138 (45.5%) were male. The number of patients increased over time, likely reflecting the wider adoption of GLP‐1RAs. Dulaglutide was the most commonly used GLP‐1RA (n = 187, 61.7%). More than half of the patients developed bowel obstruction/ileus after more than 1 year of GLP‐1RA use (n = 169, 55.8%). Approximately one‐third had a history of coronary artery disease (n = 100, 33.0%), and half had heart failure (n = 154, 50.8%). Nearly half had a history of cancer (including those diagnosed during the index hospitalization) (n = 142, 46.9%). Fifty patients (16.5%) had undergone abdominal surgery within the past 2 years, and 13 (4.3%) had a history of bowel obstruction/ileus within the past 2 years (Table 1).
TABLE 1.
Baseline characteristics of the hospitalized patients.
| Characteristics |
Patients (n = 303, %) |
|---|---|
| Age, mean (SD), years | 78.3 (9.9) |
| Age category | |
| 18–34 | 0 (0.0%) |
| 35–44 | 2 (0.7%) |
| 45–54 | 8 (2.6%) |
| 55–64 | 18 (5.9%) |
| 65–74 | 60 (19.8%) |
| 75–84 | 134 (44.2%) |
| ≥ 85 | 81 (26.7%) |
| Sex, male | 138 (45.5%) |
| Year of admission | |
| 2016 | 0 (0.0%) |
| 2017 | 7 (2.3%) |
| 2018 | 17 (5.6%) |
| 2019 | 15 (5.0%) |
| 2020 | 40 (13.2%) |
| 2021 | 60 (19.8%) |
| 2022 | 88 (29.0%) |
| 2023 | 76 (25.1%) |
| Type of health insurer | |
| Health Insurance for salaried employees (Kempo) | 6 (2.0%) |
| National Health Insurance (Kokuho) | 76 (25.1%) |
| Advanced Elderly Medical Service System | 221 (72.9%) |
| Class of GLP‐1RA | |
| Liraglutide | 63 (20.8%) |
| Exenatide | 3 (1.0%) |
| Dulaglutide | 187 (61.7%) |
| Semaglutide | 45 (14.9%) |
| Tirzepatide | 2 (0.7%) |
| Lixisenatide | 3 (1.0%) |
| Time since initiation of GLP‐1RA therapy | |
| ≥ 1 year | 169 (55.8%) |
| 7–11 months | 62 (20.5%) |
| 1–6 months | 56 (18.5%) |
| <1 month | 16 (5.3%) |
| Concomitant use of other antidiabetic drugs | |
| Insulin | 135 (44.6%) |
| Metformin | 101 (33.3%) |
| Sodium‐glucose cotransporter 2 inhibitors | 110 (36.3%) |
| Dipeptidyl peptidase‐4 inhibitors | 49 (16.2%) |
| Thiazolidines | 26 (8.6%) |
| Sulfonylureas | 65 (21.5%) |
| Alpha‐glucosidase inhibitors | 65 (21.5%) |
| Glinides | 61 (20.1%) |
| Obese or overweight | 44 (14.5%) |
| Hypertension | 208 (68.6%) |
| Dyslipidemia | 175 (57.8%) |
| Heart failure | 154 (50.8%) |
| Coronary artery disease | 100 (33.0%) |
| Stroke or transient ischemic attack | 116 (38.3%) |
| Chronic kidney disease | 103 (34.0%) |
| Dementia | 76 (25.1%) |
| Any cancer | 142 (46.9%) |
| Gastric cancer | 25 (8.3%) |
| Colonic cancer | 77 (25.4%) |
| Hepatobiliary‐pancreatic cancer | 11 (3.6%) |
| Urinary tract cancer | 14 (4.6%) |
| Gynecological cancer | 8 (2.6%) |
| Abdominal surgery within the past 2 years | 50 (16.5%) |
| Bowel obstruction/ileus within the past 2 years | 13 (4.3%) |
Abbreviations: SD, standard deviation; GLP‐1RA, glucagon‐like peptide‐1 receptor agonist.
3.2. In‐Hospital Procedures and Outcomes
A specific etiology or site of obstruction was documented in 213 (70.3%) patients, with small bowel obstruction being the most common (n = 161, 53.1%). Bowel decompression was performed in 195 (64.4%) patients, surgery was performed in 67 (22.1%) patients, and 34 (11.2%) patients required ICU admission. The median length of stay was 13 days (interquartile range, 8–24), and 18 (5.9%) patients died during hospitalization (Table 2).
TABLE 2.
Patients’ in‐hospital outcomes.
| In‐hospital outcomes |
Patients (n = 303, %) |
|---|---|
| Admission to university hospitals | 30 (9.9%) |
| Specified diagnosis of specific type of bowel obstruction/ileus | |
| Not specified | 90 (29.7%) |
| Ileus | 21 (6.9%) |
| Small bowel obstruction | 161 (53.1%) |
| Large bowel obstruction | 18 (5.9%) |
| Colonic volvulus | 13 (4.3%) |
| Bowel decompression performed | 195 (64.4%) |
| Surgery for bowel obstruction performed | 67 (22.1%) |
| Admission to the intensive care unit | 34 (11.2%) |
| Length of stay, median (IQR), days | 13 (8–24) |
| Length of stay, category | |
| ≤7 days | 67 (22.1%) |
| 8–14 days | 105 (34.7%) |
| ≥15 days | 131 (43.2%) |
| In‐hospital death | 18 (5.9%) |
Abbreviation: IQR, interquartile range.
3.3. GLP‐1RA Discontinuation and Comparisons With Other Antidiabetic Drugs
Among the 196 patients discharged alive and followed up for 6 months after discharge, 86 (43.9%) discontinued GLP‐1RAs. The discontinuation proportions for GLP‐1RAs were generally higher than those for other antidiabetic drugs, with statistically significant differences observed for insulin, metformin, and SGLT2 inhibitors (p < 0.05). In contrast, the discontinuation proportions for DPP4 inhibitors and AGIs were similar to that for GLP‐1RAs (Figure 2).
FIGURE 2.
Discontinuation proportions of GLP‐1RAs and other antidiabetic drugs. The numbers of patients who were receiving both GLP‐1RAs and the comparison drug prior to hospitalization, as well as the p values from the exact McNemar's test for statistical significance, are displayed. GLP‐1RA, glucagon‐like peptide‐1 receptor agonist; SGLT2 inhibitor, sodium‐glucose cotransporter 2 inhibitor; DPP4 inhibitor, dipeptidyl peptidase‐4 inhibitor; AGI, alpha‐glucosidase inhibitor.
3.4. Factors Associated With GLP‐1RA Discontinuation
Univariable analyses revealed that older age, shorter time since GLP‐1RA initiation, surgery for bowel obstruction, admission to the ICU, longer length of stay, and history of heart failure were significantly associated with higher discontinuation proportions (p < 0.05). No specific time trend was observed in the discontinuation proportion of GLP‐1RAs (Table 3).
TABLE 3.
Results of univariable analyses of the factors associated with GLP‐1RA discontinuation.
| Variables | GLP‐1RA discontinuation proportion | Risk difference (95% CI); p value |
|---|---|---|
| General factors | ||
| Age | ||
| 18–64 | 9/28 (32.1%) | Reference |
| 65–74 | 13/39 (33.3%) | 1.2% (–22.0% to 24.3%); p = 0.919 |
| 75–84 | 32/74 (43.2%) | 11.1% (–9.9% to 32.1%); p = 0.298 |
| ≥ 85 | 32/55 (58.2%) | 26.0% (4.0% to 48.1%); p = 0.021 |
| Sex | ||
| Male | 42/94 (44.7%) | Reference |
| Female | 44/102 (43.1%) | −1.5% (–15.6% to 12.5%); p = 0.829 |
| Year of admission | ||
| 2016–2019 | 14/31 (45.2%) | Reference |
| 2020–2021 | 38/73 (52.1%) | 6.9% (–14.3% to 28.1%); p = 0.523 |
| 2022–2023 | 34/92 (37.0%) | −8.2% (–28.6% to 12.2%); p = 0.428 |
| Type of health insurer | ||
| Health Insurance for salaried employees (Kempo) | 1/5 (20.0%) | Reference |
| National Health Insurance (Kokuho) or Advanced Elderly Medical Service System (Kouki Koureisha Iryo Seido) | 85/191 (44.5%) | 24.5% (–11.7% to 60.7%); p = 0.183 |
| Time since initiation of GLP‐1RA therapy | ||
| ≥1 year | 41/114 (36.0%) | Reference |
| 7–11 months | 21/42 (50.0%) | 14.0% (–3.8% to 31.8%); p = 0.121 |
| 1–6 months | 18/30 (60.0%) | 24.0% (4.1% to 44.0%); p = 0.018 |
| <1 month | 6/10 (60.0%) | 24.0% (–8.1% to 56.2%); p = 0.142 |
| Dementia | ||
| No | 61/150 (40.7%) | Reference |
| Yes | 25/46 (54.4%) | 13.7% (–2.9% to 30.3%); p = 0.105 |
| Any cancer | ||
| No | 43/106 (40.6%) | Reference |
| Yes | 43/90 (47.8%) | 7.2% (–6.9% to 21.3%); p = 0.314 |
| Abdominal surgery within the past 2 years | ||
| No | 69/162 (42.6%) | Reference |
| Yes | 17/34 (50.0%) | 7.4% (–11.3% to 26.1%); p = 0.435 |
| Bowel obstruction/ileus within the past 2 years | ||
| No | 81/185 (43.8%) | Reference |
| Yes | 5/11 (45.5%) | 1.7% (–29.0% to 32.3%); p = 0.914 |
| Admission to university hospitals | ||
| No | 72/173 (41.6%) | Reference |
| Yes | 14/23 (60.9%) | 19.3% (–2.2% to 40.7%); p = 0.079 |
| Factors related to the severity of bowel obstruction/ileus | ||
|---|---|---|
| Type of bowel obstruction/ileus | ||
| Not specified (reference) | 20/54 (37.0%) | Reference |
| Ileus | 5/12 (41.7%) | 4.6% (–26.7% to 36.0%); p = 0.771 |
| Small bowel obstruction | 51/108 (47.2%) | 10.2% (–6.1% to 26.5%); p = 0.218 |
| Large bowel obstruction | 5/13 (38.5%) | 1.4% (–28.6% to 31.4%); p = 0.925 |
| Colonic volvulus | 5/9 (55.6%) | 18.5% (–17.1% to 54.1%); p = 0.306 |
| Bowel decompression performed | ||
| No | 25/69 (36.2%) | Reference |
| Yes | 61/127 (48.0%) | 11.8% (–2.7% to 26.3%); p = 0.109 |
| Surgery for bowel obstruction performed | ||
| No | 53/147 (36.1%) | Reference |
| Yes | 33/49 (67.4%) | 31.3% (15.9% to 46.7%); p < 0.001 |
| Admission to the intensive care unit | ||
| No | 72/176 (40.9%) | Reference |
| Yes | 14/20 (70.0%) | 29.1% (7.5% to 50.7%); p = 0.009 |
| Length of stay | ||
| ≤7 days | 8/40 (20.0%) | Reference |
| 8–14 days | 27/73 (37.0%) | 17.0% (0.1% to 33.8%); p = 0.048 |
| ≥15 days | 51/83 (61.5%) | 41.4% (25.0% to 57.9%); p < 0.001 |
| Factors related to the potential benefits of GLP‐1RA therapy | ||
|---|---|---|
| Intensity of diabetes treatment | ||
| ≤2 types of antidiabetic drugs without insulin | 21/52 (40.4%) | Reference |
| ≥3 types of antidiabetic drugs without insulin | 25/60 (41.7%) | 1.3% (–17.2% to 19.8%); p = 0.892 |
| Treated with insulin | 40/84 (47.6%) | 7.2% (–10.1% to 24.6%); p = 0.411 |
| Obese or overweight | ||
| No | 77/171 (45.0%) | Reference |
| Yes | 9/25 (36.0%) | −9.0% (–29.5% to 11.4%); p = 0.385 |
| Hypertension | ||
| No | 26/59 (44.1%) | Reference |
| Yes | 60/137 (43.8%) | −0.3% (–15.6% to 15.1%); p = 0.972 |
| Dyslipidemia | ||
| No | 37/75 (49.3%) | Reference |
| Yes | 49/121 (40.5%) | −8.8% (–23.3% to 5.6%); p = 0.230 |
| Heart failure | ||
| No | 35/100 (35.0%) | Reference |
| Yes | 51/96 (53.1%) | 18.1% (4.3% to 32.0%); p = 0.010 |
| Coronary artery disease | ||
| No | 50/123 (40.7%) | Reference |
| Yes | 36/73 (49.3%) | 8.7% (–5.9% to 23.2%); p = 0.242 |
| Stroke or transient ischemic attack | ||
| No | 48/118 (40.7%) | Reference |
| Yes | 38/78 (48.7%) | 8.0% (–6.3% to 22.4%); p = 0.271 |
| Chronic kidney disease | ||
| No | 61/135 (45.2%) | Reference |
| Yes | 25/61 (41.0%) | −4.2% (–19.3% to 10.9%); p = 0.584 |
Abbreviations: GLP‐1RA, glucagon‐like peptide‐1 receptor agonist; CI, confidence interval.
These trends were largely preserved in the multivariable analysis. Among the general factors, older age (≥85 years vs. 18–64 years; RD = 23.0%; 95% CI 0.3% to 45.7%, p = 0.047), shorter time since GLP‐1RA initiation (for <1 month, RD = 9.3%; 95% CI ‐25.4% to 43.9%, p = 0.597; for 1–6 months, RD = 22.0%; 95% CI 2.7% to 41.3%, p = 0.025; for 7–11 months, RD = 18.6%; 95% CI 0.1% to 37.1%, p = 0.048; reference: ≥1 year), and admission to a university hospital (RD = 31.4%; 95% CI 5.7% to 57.1%, p = 0.017) were associated with higher discontinuation proportions. Among the factors related to the severity of bowel obstruction/ileus, surgery for bowel obstruction during hospitalization (RD = 32.1%; 95% CI 11.2% to 53.0%, p = 0.003), and longer length of stay (≥15 days vs. ≤7 days; RD = 32.4%; 95% CI 11.2% to 53.5%, p = 0.003) were associated with discontinuation. Among the factors related to the potential benefits of GLP‐1RA therapy, none was strongly associated with the continuation of GLP‐1RAs; conversely, heart failure was associated with discontinuation (RD = 17.6%; 95% CI 2.1% to 33.2%, p = 0.027) (Figure 3).
FIGURE 3.
Results of multivariable analysis of the factors associated with GLP‐1RA discontinuation. Point estimates and 95% CIs for each factor are displayed. *Risk difference of Health Insurance for salaried employees (Kempo) compared with National Health Insurance (Kokuho) or Advanced Elderly Medical Service System (Kouki Koureisha Iryo Seido). GLP‐1RA, glucagon‐like peptide‐1 receptor agonist; CI, confidence interval; Ref, reference.
Subgroup analyses restricted to patients undergoing surgery for bowel obstruction or those with a length of stay ≥15 days revealed that GLP‐1RAs were generally more likely to be discontinued compared with other antidiabetic drugs even within these subgroups (Figures S1 and S2).
3.5. Sensitivity Analysis
The results remained qualitatively unchanged in the sensitivity analyses, although statistical significance varied for some findings. GLP‐1RAs were generally more likely to be discontinued than other antidiabetic drugs (Figures S3 and S4). Moreover, older age, admission to university hospitals, shorter time since GLP‐1RA initiation, surgery for bowel obstruction, admission to the ICU, longer length of stay, and history of heart failure were associated with higher discontinuation proportions of GLP‐1RAs (Figures S5 and S6).
4. Discussion
The study population was characterized by advanced age (mean 78) and a relatively high burden of comorbidities. Within 6 months of discharge, 43.9% of patients discontinued GLP‐1RAs, and the discontinuation proportion was generally higher than that for other antidiabetic drugs, except for DPP4 inhibitors and AGIs. Older age and shorter time since GLP‐1RA initiation were associated with discontinuation, as were some factors related to the severity of bowel obstruction/ileus (surgery for bowel obstruction and longer length of stay). In contrast, none of the factors related to the potential benefits of GLP‐1RAs was associated with continuation.
The discontinuation proportion of 43.9% observed at 6 months in this study is high, although it aligns with previous real‐world studies in general patients with type 2 diabetes or obesity, which reported 6–12‐month discontinuation proportions of approximately 15%–50% [29, 30, 31, 32]. The higher discontinuation proportion compared with other antidiabetic drugs such as insulin, metformin, and SGLT2 inhibitors observed in our study may reflect physicians’ awareness of, and responsiveness to, regulatory safety warnings regarding GLP‐1RAs and bowel obstruction/ileus. Previous reports have shown that such regulatory safety warnings often influence physicians’ prescribing behavior [33]. Even in the absence of such specific awareness, the well‐known gastrointestinal side‐effect profile of GLP‐1RAs [34] may have led physicians to discontinue them after the bowel obstruction/ileus event. Moreover, the finding that the discontinuation proportion for GLP‐1RAs was similar to those of DPP4 inhibitors and AGIs is plausible, as there have been case reports and general concerns about gastrointestinal side‐effects, including bowel obstruction/ileus, associated with these drugs [35, 36].
The finding that older patients were more likely to discontinue GLP‐1RAs is consistent with that of several previous studies investigating factors associated with discontinuation in general patients with type 2 diabetes or obesity [30, 31, 32]. Several possible explanations can be posited. First, physicians may have been more cautious about avoiding the recurrence of bowel obstruction/ileus in older patients. Second, prescription optimization may have been undertaken during hospitalization [37], leading to discontinuation of medications including GLP‐1RAs in older patients with polypharmacy [38]. Third, the bowel obstruction/ileus event may have been more likely to result in reduced physical function or appetite in older patients, diminishing the need for glucose‐lowering therapy. The association between a shorter duration since GLP‐1RA initiation and discontinuation may also reflect the perception that the event was drug‐related.
A higher severity of events (surgery, longer length of stay) was associated with higher discontinuation proportions. If physicians attributed the event to GLP‐1RA use, discontinuation after more severe events is understandable, because they would be more inclined to avoid recurrence in such cases. While it remains possible that severe events reduced the need for glucose‐lowering drugs (e.g., due to reduced oral intake), this explanation alone is less likely, as GLP‐1RAs were still generally more frequently discontinued than other antidiabetic drugs in these subgroups.
Patients who potentially derive greater benefits from GLP‐1RAs, such as those with obesity or overweight, cardiovascular disease, or chronic kidney disease, did not have higher continuation proportions; in fact, GLP‐1RAs were more likely to be discontinued in patients with heart failure. These findings align with prior research in the general population that found no association or even a higher likelihood of discontinuation in patients with cardiovascular risk factors [30, 31]. Patients with heart failure may be perceived as more vulnerable to adverse events [39], which may have led physicians to prioritize safety over the potential benefits of GLP‐1RAs.
To summarize the discussion above, our findings suggest that physicians often perceived GLP‐1RAs as potential contributors to the occurrence of bowel obstruction/ileus, possibly influenced by regulatory safety warnings, and were consequently more likely to discontinue them after severe events. Conversely, the potential therapeutic benefits of GLP‐1RAs appeared to play a smaller role in continuation decisions. Given the uncertainty about whether GLP‐1RAs truly increase the risk of bowel obstruction/ileus, such decision‐making may not always be justified. Patients who are likely to benefit more from GLP‐1RA therapy (e.g., those with obesity, cardiovascular disease, or chronic kidney disease) may derive a greater net benefit from continuation. However, to support a more balanced risk–benefit‐based decision‐making, further studies are needed to clarify whether GLP‐1RAs truly increase the risk of bowel obstruction/ileus, and, if so, in which specific patient groups.
This study has several limitations. First, the relatively small sample size limited statistical power, yielding generally wide confidence intervals. Consequently, the absence of statistical significance does not necessarily rule out clinically meaningful associations. Second, we lacked data on the direct reasons for discontinuation, and on potentially influential factors such as daily gastrointestinal symptoms during treatment, detailed weight or glycemic changes [40], or household income [30]. Third, not all ICD‐10 code‐based diagnoses in Japanese claim data have been validated. We attempted to improve diagnostic accuracy by combining diagnostic codes with prescriptions or procedures. Fourth, health check‐up data (body mass index, estimated glomerular filtration rate, and proteinuria) were available for only 10%–17% of patients, which may have led to underestimation of the prevalence of obesity/overweight and chronic kidney disease. Fifth, while bowel obstruction is explicitly listed as a serious adverse drug reaction in Japan for some GLP‐1RAs [6, 7, 8], it is listed only as a potential adverse event in the United States [41] and Europe [42]. Therefore, our findings may not be directly generalizable to countries outside Japan. Finally, we did not evaluate whether continuation versus discontinuation of GLP‐1RA therapy after bowel obstruction/ileus events was associated with subsequent clinical outcomes because of the limited sample size.
In conclusion, this nationwide claims‐based study found that 43.9% of patients discontinued GLP‐1RAs at 6 months after a bowel obstruction/ileus event. Older age, shorter duration since GLP‐1RA initiation, surgery for bowel obstruction, longer length of stay, and heart failure were associated with higher discontinuation proportions. While GLP‐1RAs were more likely to be discontinued after severe events, their potential therapeutic benefits appeared less frequently considered in continuation decisions.
Funding
This work was supported by the Ministry of Health, Labour and Welfare (23AA2003)
Conflicts of Interest
The authors declare no conflicts of interest.
Supporting information
Figure S1: Discontinuation proportions of GLP‐1RAs and other antidiabetic drugs among patients who underwent surgery for bowel obstruction during hospitalization.
Figure S2: Discontinuation proportions of GLP‐1RAs and other antidiabetic drugs among patients with a length of stay ≥15 days.
Figure S3: Discontinuation proportions of GLP‐1RAs and other antidiabetic drugs: hospitalization for bowel obstruction/ileus defined on the basis of radiography or computed tomography performed on the day of admission.
Figure S4: Discontinuation proportions of GLP‐1RAs and other antidiabetic drugs: including patients who could not be followed‐up for 6 months after discharge.
Figure S5: Results of multivariable analysis of the factors associated with GLP‐1RA discontinuation: hospitalization for bowel obstruction/ileus defined on the basis of radiography or computed tomography performed on the day of admission.
Figure S6: Results of multivariable analysis of the factors associated with GLP‐1RA discontinuation: including patients who could not be followed‐up for 6 months after discharge.
References
- 1. Sattar N., Lee M. M. Y., Kristensen S. L., et al., “Cardiovascular, Mortality, and Kidney Outcomes With GLP‐1 Receptor Agonists in Patients With Type 2 Diabetes: A Systematic Review and Meta‐Analysis of Randomised Trials,” Lancet Diabetes & Endocrinology 9, no. 10 (2021): 653–662. [DOI] [PubMed] [Google Scholar]
- 2. Yao H., Zhang A., Li D., et al., “Comparative Effectiveness of GLP‐1 Receptor Agonists on Glycaemic Control, Body Weight, and Lipid Profile for Type 2 Diabetes: Systematic Review and Network Meta‐analysis,” Bmj 384 (2024): e076410. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 3. Pharmaceuticals and Medical Devices Agency , “Pharmaceuticals and Medical Devices Safety Information,” No. 291, June 2012, Accessed June 2025, https://www.pmda.go.jp/files/000153459.pdf.
- 4. European Medicines Agency , “Pharmacovigilance Risk Assessment Committee (PRAC),” Minutes of the Meeting —7–10 January 2013, Accessed June 2025, https://www.ema.europa.eu/en/documents/minutes/minutes‐prac‐meeting‐7‐10‐january‐2013_en.pdf.
- 5. U.S. Food and Drug Administration , “Potential Signals of Serious Risks/New Safety Information Identified by the FDA Adverse Event Reporting System (FAERS),” April–June 2022, Accessed June 2025, https://www.fda.gov/drugs/fdas‐adverse‐event‐reporting‐system‐faers/april‐june‐2022‐potential‐signals‐serious‐risksnew‐safety‐information‐identified‐fda‐adverse‐event.
- 6.“Product Information: Victoza Subcutaneous Injection (liraglutide),” Accessed July 2025, https://www.pmda.go.jp/PmdaSearch/iyakuDetail/ResultDataSetPDF/620023_2499410G1021_1_12, Updated February 2024.
- 7.“Product Information: Byetta Subcutaneous Injection Pen (Exenatide),” Accessed July 2025, https://www.pmda.go.jp/PmdaSearch/iyakuDetail/ResultDataSetPDF/670227_2499411G1026_2_07, Updated February 2023.
- 8.“Product Information: Trulicity Subcutaneous Injection Ateos (Dulaglutide),” Accessed July 2025, https://www.pmda.go.jp/PmdaSearch/iyakuDetail/ResultDataSetPDF/530471_2499416G1029_1_20, Updated June 2025.
- 9. Faillie J. L., Yin H., Yu O. H. Y., et al., “Incretin‐Based Drugs and Risk of Intestinal Obstruction among Patients with Type 2 Diabetes,” Clinical Pharmacology & Therapeutics 111, no. 1 (2022): 272–282. [DOI] [PubMed] [Google Scholar]
- 10. Ueda P., Wintzell V., Melbye M., et al., “Use of DPP4 Inhibitors and GLP‐1 Receptor Agonists and Risk of Intestinal Obstruction: Scandinavian Cohort Study,” Clinical Gastroenterology and Hepatology 22, no. 6 (2024): 1226–1237.e14. [DOI] [PubMed] [Google Scholar]
- 11. Gao Z., Tabernacki T., Dorney I., Ding P., Kaelber D. C., and Xu R., “Association of GLP‐1 Receptor Agonists With Risk of Intestinal Obstruction in Patients With Type 2 Diabetes Mellitus: A Retrospective Cohort Study,” Acta Diabetologica 62, no. 11 (2025): 1941–1951. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 12. Sarwal A., Singh R., Wei G., et al., “Glycemic Therapies and the Risk of Gastrointestinal Adverse Events in Veterans With Type 2 Diabetes,” Diabetes, Obesity and Metabolism 27, no. 10 (2025): 5865–5877. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 13. Chiang C. H., Jaroenlapnopparat A., Colak S. C., et al., “Glucagon‐Like Peptide‐1 Receptor Agonists and Gastrointestinal Adverse Events: A Systematic Review and Meta‐Analysis,” Gastroenterology 169, no. 6 (2025): 1268–1281. [DOI] [PubMed] [Google Scholar]
- 14. Edwards I. R. and Aronson J. K., “Adverse Drug Reactions: Definitions, Diagnosis, and Management,” Lancet 356, no. 9237 (2000): 1255–1259. [DOI] [PubMed] [Google Scholar]
- 15. Coleman J. J. and Pontefract S. K., “Adverse Drug Reactions,” Clinical Medicine (London) 16, no. 5 (2016): 481–485. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 16. Brahmer J. R., Lacchetti C., Schneider B. J., et al., “Management of Immune‐Related Adverse Events in Patients Treated with Immune Checkpoint Inhibitor Therapy,” Journal of Clinical Oncology 36, no. 17 (2018): 1714–1768. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 17. Khan D. A., Banerji A., Blumenthal K. G., et al., “Drug Allergy: A 2022 Practice Parameter Update,” Journal of Allergy and Clinical Immunology 150, no. 6 (2022): 1333–1393. [DOI] [PubMed] [Google Scholar]
- 18. Singh K. P., Carvalho A. C. C., Centis R., et al., “Clinical Standards for the Management of Adverse Effects During Treatment for TB,” International Journal of Tuberculosis and Lung Disease 27, no. 7 (2023): 506–519. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 19. Jung J. W., Kang H. R., and Kim M. H., “Immediate Hypersensitivity Reaction to Gadolinium‐Based MR Contrast media,” Radiology 264, no. 2 (2012): 414–422. [DOI] [PubMed] [Google Scholar]
- 20. Zanotti K. M. and Markman M., “Prevention and Management of Antineoplastic‐Induced Hypersensitivity Reactions,” Drug Safety 24, no. 10 (2001): 767–779. [DOI] [PubMed] [Google Scholar]
- 21. Morgenthaler L. and DePietro R., “Case Report: Small Bowel Obstruction After Starting Tirzepatide (Mounjaro),” American Family Physician 110, no. 6 (2024): 562. [PubMed] [Google Scholar]
- 22. Jones M. and Cappola J. J., “GLP‐1 Receptor Agonists in Diabetes and Obesity: A Case Report and Review of Bowel Obstruction Risks and Management,” Cureus 17, no. 4 (2025): e81891. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 23. Wilding J. P. H., Batterham R. L., Davies M., et al., “Weight Regain and Cardiometabolic Effects After Withdrawal of Semaglutide: The STEP 1 Trial Extension,” Diabetes, Obesity and Metabolism 24, no. 8 (2022): 1553–1564. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 24. Rubino D., Abrahamsson N., Davies M., et al., “Effect of Continued Weekly Subcutaneous Semaglutide vs Placebo on Weight Loss Maintenance in Adults With Overweight or Obesity: The STEP 4 Randomized Clinical Trial,” Jama 325, no. 14 (2021): 1414–1425. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 25. Piccini S., Favacchio G., Panico C., et al., “Time‐Dependent Effect of GLP‐1 Receptor Agonists on Cardiovascular Benefits: A Real‐World Study,” Cardiovascular Diabetology 22, no. 1 (2023): 69. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 26. Yasunaga H., “DeSC Database,” Annals of Epidemiology 7, no. 2 (2025): 46–49. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 27. Okada A. and Yasunaga H., “Prevalence of Noncommunicable Diseases in Japan Using a Newly Developed Administrative Claims Database Covering Young, Middle‐Aged, and Elderly People,” Jma Journal 5, no. 2 (2022): 190–198. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 28. Cheung Y. B., “A Modified Least‐Squares Regression Approach to the Estimation of Risk Difference,” American Journal of Epidemiology 166, no. 11 (2007): 1337–1344. [DOI] [PubMed] [Google Scholar]
- 29. Do D., Lee T., Peasah S. K., Good C. B., Inneh A., and Patel U., “GLP‐1 Receptor Agonist Discontinuation among Patients With Obesity and/or Type 2 Diabetes,” Jama Network Open 7, no. 5 (2024): e2413172. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 30. Lassen M. C. H., Johansen N. D., Modin D., et al., “Adherence to Glucagon‐Like Peptide‐1 Receptor Agonist Treatment in Type 2 Diabetes Mellitus: A Nationwide Registry Study,” Diabetes, Obesity and Metabolism 26, no. 11 (2024): 5239–5250. [DOI] [PubMed] [Google Scholar]
- 31. Rodriguez P. J., Zhang V., Gratzl S., Do D., and Baker C., “Discontinuation and Reinitiation of Dual‐Labeled GLP‐1 Receptor Agonists Among US Adults With Overweight or Obesity,” Jama Network Open 8, no. 1 (2025): e2457349. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 32. Weiss T., Carr R. D., Pal S., et al., “Real‐World Adherence and Discontinuation of Glucagon‐Like Peptide‐1 Receptor Agonists Therapy in Type 2 Diabetes Mellitus Patients in the United States,” Patient Prefer Adherence 14 (2020): 2337–2345. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 33. Georgi U., Lämmel J., Datzmann T., Schmitt J., and Deckert S., “Do Drug‐Related Safety Warnings Have the Expected Impact on Drug Therapy? A Systematic Review,” Pharmacoepidemiology and Drug Safety 29, no. 3 (2020): 229–251. [DOI] [PubMed] [Google Scholar]
- 34. Gorgojo‐Martinez J. J., Mezquita‐Raya P., Carretero‐Gomez J., et al., “Clinical Recommendations to Manage Gastrointestinal Adverse Events in Patients Treated With Glp‐1 Receptor Agonists: A Multidisciplinary Expert Consensus,” Journal of Clinical Medicine 12, no. 1 (2022): 145. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 35. Kanasaki K., Konishi K., Hayashi R., et al., “Three ileus Cases Associated With the Use of Dipeptidyl Peptidase‐4 Inhibitors in Diabetic Patients,” Journal of Diabetes Investigation 4, no. 6 (2013): 673–675. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 36. Oba K., Suzuki K., Ouchi M., Matsumura N., Suzuki T., and Nakano H., “Repeated Episodes of Paralytic Ileus in an Elderly Diabetic Patient Treated With voglibose,” Journal of the American Geriatrics Society 54, no. 1 (2006): 182–183. [DOI] [PubMed] [Google Scholar]
- 37. Weiss O., Eyre A., Ellenbogen D. A., and Stein G. Y., “The Impact of Hospitalization on Inappropriate Prescribing and Polypharmacy in Older Patients: A Descriptive Cross‐Sectional Study,” Pharmacoepidemiology and Drug Safety 33, no. 5 (2024): e5812. [DOI] [PubMed] [Google Scholar]
- 38. Hamada S., Iwagami M., Sakata N., et al., “Changes in Polypharmacy and Potentially Inappropriate Medications in Homebound Older Adults in Japan, 2015–2019: A Nationwide Study,” Journal of General Internal Medicine 38, no. 16 (2023): 3517–3525. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 39. Pandey A., Kitzman D., and Reeves G., “Frailty Is Intertwined With Heart Failure: Mechanisms, Prevalence, Prognosis, Assessment, and Management,” Jacc Heart Failure 7, no. 12 (2019): 1001–1011. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 40. Sikirica M. V., Martin A. A., Wood R., Leith A., Piercy J., and Higgins V., “Reasons for Discontinuation of GLP1 Receptor Agonists: Data From a Real‐world Cross‐Sectional Survey of Physicians and Their Patients With Type 2 Diabetes,” Diabetes, Metabolic Syndrome and Obesity 10 (2017): 403–412. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 41.“Highlights of Prescribing Information: Victoza (Liraglutide) Injection, for Subcutaneous Use,” Accessed August 2025, https://www.accessdata.fda.gov/drugsatfda_docs/label/2025/022341s044lbl.pdf. Updated May 2025.
- 42.“Victoza: EPARProduct Information. European Medicines Agency,” Accessed June 2025, https://www.ema.europa.eu/en/documents/product‐information/victoza‐epar‐product‐information_en.pdf, Updated February 2025.
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.
Supplementary Materials
Figure S1: Discontinuation proportions of GLP‐1RAs and other antidiabetic drugs among patients who underwent surgery for bowel obstruction during hospitalization.
Figure S2: Discontinuation proportions of GLP‐1RAs and other antidiabetic drugs among patients with a length of stay ≥15 days.
Figure S3: Discontinuation proportions of GLP‐1RAs and other antidiabetic drugs: hospitalization for bowel obstruction/ileus defined on the basis of radiography or computed tomography performed on the day of admission.
Figure S4: Discontinuation proportions of GLP‐1RAs and other antidiabetic drugs: including patients who could not be followed‐up for 6 months after discharge.
Figure S5: Results of multivariable analysis of the factors associated with GLP‐1RA discontinuation: hospitalization for bowel obstruction/ileus defined on the basis of radiography or computed tomography performed on the day of admission.
Figure S6: Results of multivariable analysis of the factors associated with GLP‐1RA discontinuation: including patients who could not be followed‐up for 6 months after discharge.