Health‐care resource use and costs associated with diabetic and idiopathic gastroparesis: A claims analysis of the first 3 years following the diagnosis of gastroparesis

Abstract Background Due to limited treatment options, many patients with diabetic gastroparesis (DG) or idiopathic gastroparesis (IG) experience inadequate symptom control resulting in increased health‐care resource utilization (HRU) and associated costs. We compared all‐cause HRU and health‐care costs over the 3 years after patients’ first gastroparesis diagnosis with that of matched controls without gastroparesis. Methods Newly diagnosed adults with DG or IG were identified in Optum's de‐identified Clinformatics® Data Mart Database (Q1‐2007 to Q1‐2019). Patients with DG/IG were matched 1:1 to controls using a mixed approach of exact matching and propensity score matching. The index date was the first gastroparesis diagnosis for cases or randomly selected for controls. All‐cause HRU and direct health‐care costs per person‐year (PPY) were compared between DG/IG cases and controls in Years 1–3 post‐index. Key Results Demographics and comorbidities were balanced between patients with gastroparesis (n = 18,015 [DG]; n = 14,305 [IG]) and controls. In each of the Years 1–3 post‐index, patients with DG or IG had significantly higher annual HRU and costs versus controls (mean total cost differences PPY: DG Year 1 $34,885, Year 2 $28,071, Year 3 $25,606; IG Year 1 $23,176, Year 2 $16,627, Year 3 $14,396) (all p < 0.05). Across all 3 years, DG/IG cohorts had approximately twice the costs of controls. HRU and costs were highest in Year 1 post‐index for both DG and IG. Conclusions & Inferences The economic burden of gastroparesis remains high several years after diagnosis, emphasizing the need for chronic treatment to effectively manage symptoms and consequently reduce the burden of this disorder.


| INTRODUC TI ON
Gastroparesis is a chronic disorder characterized by reduced stomach muscle motility and delayed gastric emptying without mechanical obstruction. 1 The most commonly reported subtypes are idiopathic gastroparesis (IG) and diabetic gastroparesis (DG). 2

,3 IG
represents a prevalent etiology subgroup in which a cause cannot be identified. 4,5 DG is often associated with poorly controlled type I or type II diabetes and may occur with other diabetic complications and result in frequent hospitalizations. [6][7][8] Regardless of type, symptoms of gastroparesis include nausea, vomiting, early satiety, bloating, upper abdominal pain, and post-prandial fullness, which may negatively impact nutrition. 3 In severe cases, patients may experience weight loss, malnutrition, dehydration, electrolyte imbalance, hypoor hyperglycemic fluctuations among diabetics, and bezoar formation. 3,9 The overall prevalence of gastroparesis in the United States (US) was estimated at 16 per 100,000 persons during 1999-2014. 10 While mild gastroparesis may be managed with nonpharmacologic measures (e.g., dietary modification and hydration), patients with prolonged or moderate-to-severe gastroparesis symptoms are often managed with medications (e.g., prokinetics to improve gastric motility and antiemetics to reduce vomiting and nausea). [11][12][13] The prokinetic metoclopramide is the only currently approved agent by the US Food and Drug Administration for the treatment of gastroparesis (DG only), but it carries a black box warning for tardive dyskinesia and other extrapyramidal symptoms. 13 No medication is approved for gastroparesis in the European Union. Furthermore, other prokinetics or antiemetic treatment options have limited efficacy, and some are associated with side effects such as cardiovascular and extrapyramidal events. 14,15 Gastroparesis is also associated with a high economic burden and a negative impact on health-related quality of life (HRQoL), [16][17][18][19] likely related to the lack of effective treatments for persistent, uncontrolled symptoms, and delayed diagnosis. Indeed, prior studies indicate that delays in diagnosis or misdiagnosis of gastroparesis are common in real-world practice due to the overlap between its symptoms and those of other gastrointestinal disorders (e.g., functional dyspepsia and ulcer). [20][21][22][23][24] Furthermore, patients with gastroparesis often have lower HRQoL with limited ability to perform daily activities due to uncontrolled symptoms. 17,25 Wadhwa et al. found that the number of inpatient admissions due to gastroparesis had risen significantly over time, contributing estimated costs of over $550 million in 2017 to the US health system. 16 However, real-world evidence on the health-care resource utilization (HRU) and costs associated with gastroparesis remains limited, especially long-term HRU and costs after the diagnosis of gastroparesis. Given the lack of effective treatments for gastroparesis, 14,15 the associated economic burden may be expected to manifest long term. Furthermore, prior studies have primarily focused on inpatient hospital admissions or emergency room (ER) visits and have not used matched patients without gastroparesis as controls to assess the additional costs and HRU associated with gastroparesis. 16,17,26 To address this knowledge gap of the real-world HRU and costs with gastroparesis, this study used a large US administrative claims database to quantify the direct medical burden of gastroparesis (allcause HRU and associated costs) among patients with DG or IG in the first 3 years following the initial gastroparesis diagnosis compared to matched controls without gastroparesis.

| Data source
This study used de-identified data from the Optum Clinformatics This study used anonymized claims data; thus, no institutional board review was required.

| Selection of DG and IG cohorts and matched controls
Adults (aged ≥18 years) newly diagnosed with DG or IG between

Key Points
• The aim of this retrospective US claims study was to quantify the direct medical burden of gastroparesis (allcause healthcare resource utilization and associated costs attributable to having gastroparesis).
• This study demonstrates the substantial economic burden associated with diabatic and idiopathic gastroparesis over each of the first three years after the initial diagnosis of gastroparesis.
• The results emphasize the need for chronic treatment to effectively manage symptoms and consequently reduce the burden of this disorder. F I G U R E 1 Sample Selection of (A) the DG Cohort and Matched Controls and (B) the IG Cohort and Matched Controls. Abbreviations: DG, diabetic gastroparesis; IG, idiopathic gastroparesis; Q, quarter. 1 Diagnosis claims for gastroparesis were required to be incurred in an inpatient facility, an outpatient facility, or from professional services (excluding laboratory, radiology, and pathology) settings. 2 For patients with only one gastroparesis claim, the gastric emptying procedure was required to be before the date of the gastroparesis claim. 3 DG and IG potential controls were matched 1:1 to patients in the respective cohort using a mixed approach of exact matching on some covariates (age at index date, sex, and length of continuous enrollment) and propensity score matching ≥2 years and ≥3 years of continuous enrollment after the index date were included in the analyses for 2-year and 3-year followup, respectively.
Patients newly diagnosed with gastroparesis who met the selection criteria (as shown in Figure 1A,B) were categorized into two mutually exclusive cohorts. The DG cohort included those with at least one diagnosis claim for type I or type II diabetes before the index date. The IG cohort included the remaining patients after excluding those with fundoplication or with a diagnosis of Parkinson's disease before the index date.

| Matching of controls with the DG and IG cohorts
Potential controls for the DG and IG cohorts were selected from the general population using similar criteria as those used for the DG and IG cohorts ( Figure 1). For the potential controls, the index date was selected randomly from all dates with ≥1 medical claim. Controls for the DG cohort were required to have ≥1 diagnosis of type I or type II diabetes before their index date. DG controls and IG controls were matched 1:1 to the patients in the DG or IG cohorts, respectively, using a mixed approach of exact matching on key covariates (age at index date, sex, and years of continuous enrollment; same value required) and matching on propensity score using a 0.01 caliper (i.e., the maximum difference between matched subjects' propensity scores).
For both cohorts, the propensity scores used for matching were derived from logistic models that included demographics (age, sex, race, payer type, and region of residence), socioeconomic characteristics (income and education), year of the index date, and all individual comorbidities included in the Charlson Comorbidity Index (CCI).

| Measurements
Patient characteristics, measured during the baseline period, included all matching variables in the propensity score models and additional variables that may reflect the burden of gastroparesis pre-diagnosis (Table 1). Comorbidities more commonly observed among patients with gastroparesis were identified according to the list reported by Nassar and Richter. 32 All-cause HRU outcomes included inpatient admissions and total inpatient days, and ER, outpatient, and other health-care use (e.g., durable medical equipment, transportation services, and other professional services such as diagnostic testing, laboratory, or radiology). All-cause total cost outcomes included all-cause costs (i.e., estimated claims paid) for pharmacy and medical services (inpatient, ER, outpatient, and other visits). Health-care costs were adjusted to 2019 US dollars (USD) using the medical care component of the Consumer Price Index. 33

| Statistical analysis
Baseline characteristics including demographics as of the index date and comorbidities, HRU, and costs in the year prior to the index date were described using means, standard deviations (SD), frequencies, and percentages. For patient characteristics in the propensity score models, standardized mean differences between patients with DG/IG and their respective matched controls were used to evaluate the matching performance (typically, absolute value of standardized differences within 10% is perceived as balanced). 34  This study compared the all-cause HRU and associated costs among patients with a gastroparesis diagnosis to that of the matched controls from the general population, and the difference in the burden was assumed attributable to gastroparesis. All-cause HRU Over the up to 3-year follow-up period, HRU incidence rate ratios (IRR) were estimated to compare the incidence rates of admission/visit occurrence using generalized estimating equations (GEE) models with binomial distribution and repeated HRU measurements (one HRU measurement for each year with complete follow-up postindex). GEE models with Tweedie distribution were used to estimate cost differences over the follow-up period. In both cases, the GEE methodology accounted for the repeated measures within subjects over the 3-year follow-up and for the matched design.

| Sensitivity analyses
In the sensitivity analyses, the regression models for HRU and costs were further adjusted for comorbidities more commonly seen in patients with gastroparesis 32 (unmatched variables), total all-cause baseline costs (unmatched variables), and index year (matched variable, but adjusted in the model because the standardized difference was >|0.10| post-matching). Because this approach further removes the pre-index differences between cohorts that may reflect the burden of gastroparesis pre-diagnosis (e.g., anemia, functional dyspepsia, or anxiety disorder 32 ), the estimates are expected to be conservatively low.

| Patients with DG versus matched diabetic controls
After meeting all eligibility criteria, 18,015 patients with DG were matched to diabetic controls (i.e., without gastroparesis and with ≥1 year of follow-up post-index) and included in the study ( Figure 1A).

| Baseline characteristics
Among patients with DG, the mean age was 62 years, 66% were female, and 55% were white (

| Patients with IG versus matched nondiabetic controls
After meeting all eligibility criteria, 14,305 IG-matched pairs were included in the study ( Figure 1B). By design, all of these patients had at least 1 year of follow-up post-index.
Health-care costs measured during each of the 3 years of follow-up were all significantly higher among the IG cohort compared with matched controls (Figure 3A-C

| Sensitivity analysis
The findings from the sensitivity analysis were consistent with those of the main analyses (

| DISCUSS ION
Gastroparesis may be associated with debilitating symptoms, but   Prior analyses of the economic burden of gastroparesis have generally focused on limited health-care settings, primarily inpatient, or used data other than claims. 16 which may lead to more frequent follow-up visits during that process. Another possibility is that the patients with severe symptoms or the very ill-who would be likely to have the costliest HRU-were lost to follow-up in Years 2 and 3 of the analysis. However, although the HRU and costs were lower in Years 2 and 3 for the DG and IG cohorts, the overall burden of gastroparesis remained twice as high in those years compared to matched controls. This consistently high economic burden of gastroparesis after initial diagnosis and treatment could be related to the lack of effective treatment options that safely manage its chronic symptoms. 14,15 The suboptimal management options for gastroparesis symptoms may result in short treatment durations and frequent therapy switching. A recent study found that patients newly diagnosed with gastroparesis often remain untreated for extended periods after an initial attempt with conventional therapies, including metoclopramide and other treatments used in real-world practice. 35 In general, treatment duration is brief during the 1st year following gastroparesis diagnosis and 39%-57% of patients switched therapies in the first few months, 35 indicating poor symptom control or tolerability issues. Indeed, 60% of patients with gastroparesis in a 2017 survey expressed dissatisfaction with their therapy. 25 The burden of persistent gastroparesis symptoms 16,36-38 and its negative impact on HRQoL and work productivity, 17 as described in prior studies, also indicate the suboptimal performance of current treatments.
Multiple studies have reported clinically assessed prevalence of persistent gastroparesis symptoms, including nausea and vomiting (up to 58% of patients), 36 abdominal pain (up to 51%), 16,36,38 and diarrhea and constipation (around 17%). 37 Other studies of self-reported outcomes have confirmed the high symptom burden of gastroparesis, with 85-95% of patients reporting nausea and 25-45% reporting abdominal pain, [39][40][41][42] and high rates of total parenteral nutrition (~20% in the prior year) or a feeding tube (19%), 17 and reduction in daily activities, school/work schedules, and annual income. 17 Patients with gastroparesis also experience frequent 30-and 90-day inpatient readmissions, estimated at 26.8-35% and 45.6%, respectively. 43,44 The indirect costs of gastroparesis, and the subsequent loss of income or opportunity, further contribute to the overall disease burden to patients and society. Furthermore, a rising incidence of gastroparesis diagnoses (both IG and DG), and hospitalizations and ER visits related to gastroparesis, has been reported over different time periods from 1997 to 2013. 16,26 This rise in incidence raises concern for escalating costs to the health-care system related to gastroparesis. This study is also subject to a few limitations. First, as with any retrospective claims database, occasional miscoding may occur in the claims data. Furthermore, as patients may change insurance plans over years, it could be possible that some patients were previously diagnosed with gastroparesis but not reflected in this claims database. Second, gastroparesis is often underdiagnosed (e.g., with delay) or misdiagnosed (e.g., as functional dyspepsia 23 ) in real-world practice. This may result in the underestimation of the burden of gastroparesis in this study and also may explain the high prevalence of baseline comorbidities related to gastroparesis among newly diagnosed patients. Thus, a portion of the burden of gastroparesis (i.e., diagnostic tests and procedures) may occur prior to the initial diagnosis, which was managed via analytical methods by assessing costs and HRU during the 1 year prior to the index date (baseline period). The sensitivity analyses adjusting for additional baseline comorbidities and health-care costs provided an estimate which is likely lower than the true burden of gastroparesis. Third, these results primarily apply to patients covered by commercial health insurance and may not represent patients with other types of insurance or the uninsured. Fourth, the parameters used when defining the IG cohort, which is the residual category after the assignment of other etiologies, may not be exhaustive. Fifth, there are several limitations related to the identification of patients with diabetes using the Optum claims data. Type I and type II diabetes were identified in the database using diagnosis codes, which could be subject to potential misclassification and coding errors. We classified approximately 25.9% of DG patients with unknown diabetes type because they had codes for both type I and type II diabetes during the baseline period. In addition, insulin use was calculated based only on the 1-year baseline period and among the patients with diabetes who met the sample selection criteria, which may undercapture patients' insulin use history. The relative risks of the occurrence of diabetic complications by type I or type II diabetes were not assessed in this economic burden study. Lastly, the results may include bias due to unobserved factors that could not be adjusted for as covariates in the matched design.
In conclusion, this retrospective US claims study demonstrates the substantial economic burden associated with DG and IG over each of the first 3 years after the initial diagnosis of gastroparesis.
This increased economic burden is reflected by patients' greater comorbidity burden and higher all-cause HRU and health-care costs compared with matched controls. The matched design permitted the ability to determine the difference in HRU and costs attributable to gastroparesis. The difference in the economic burden experienced by patients with DG or IG versus their matched controls was the highest in the 1st year after gastroparesis diagnosis and remained significantly higher than that of controls over all 3 years of follow-up.
These findings emphasize the need for gastroparesis therapies suitable for chronic symptom and disease management, which may reduce the burden of illness and improve the HRQoL of patients with gastroparesis.

ACK N OWLED G M ENTS
Medical writing assistance was provided by Shelley Batts, PhD, an employee of Analysis Group, Inc. Support for this assistance was provided by Takeda Pharmaceuticals.

AUTH O R CO NTR I B UTI O N S
All authors designed the research study, performed the research, and contributed to drafting and editing the paper. WT, RA, RI, and EW analyzed the data.