Epidemiology, Comorbidities, and Treatment of Cyclic Vomiting Syndrome in the United States

Abstract

INTRODUCTION:

Cyclic vomiting syndrome (CVS) imposes a substantial burden, but epidemiological data are scarce. This study aimed to estimate the incidence and prevalence of CVS, comorbid conditions, and treatment patterns, using administrative databases in the United States.

METHODS:

This cross-sectional study used claims data from Merative MarketScan Commercial/Medicare Supplemental and Medicaid databases in all health care settings. Incidence and prevalence rates for 2019 were calculated and stratified by age, sex, region, and race/ethnicity. Patient characteristics were reported among newly diagnosed patients with CVS (i.e., no documented claims for CVS before 2019). CVS was defined as having 1+ inpatient and/or 2+ outpatient CVS claims that were 7+ days apart.

RESULTS:

The estimated prevalence of CVS was 16.7 (Commercial/Medicare) and 42.9 (Medicaid) per 100,000 individuals. The incidence of CVS was estimated to be 10.6 (Commercial/Medicare) and 26.6 (Medicaid) per 100,000 individuals. Both prevalence and incidence rates were higher among female individuals (for both Commercial/Medicare and Medicaid). Comorbid conditions were common and included abdominal pain (56%–64%), anxiety (32%–39%), depression (26%–34%), cardiac conditions (39%–42%), and gastroesophageal reflux disease (30%–40%). Despite a diagnosis of CVS, only 32%–35% had prescriptions for prophylactic treatment and 47%–55% for acute treatment within the first 30-day period following diagnosis.

DISCUSSION:

This study provides the first population-level estimates of CVS incidence and prevalence in the United States. Comorbid conditions are common, and most patients with CVS do not receive adequate treatment. These findings underscore the need for improving disease awareness and developing better screening strategies and effective treatments.

INTRODUCTION

Cyclic vomiting syndrome (CVS) is a poorly understood disorder of gut-brain interaction, characterized by sudden episodes of severe nausea and vomiting (1–4). Vomiting may occur multiple times an hour, with episodes lasting hours to several days (2). Patients often require repeated emergency department visits or hospitalizations due to the recalcitrant nature of their symptoms (1). Patients with CVS often also have multiple comorbidities such as anxiety, depression, autonomic dysfunction, and migraine, which complicate management, negatively affect patient outcomes, and reduce overall quality of life (1,2,5,6). Unfortunately, despite the substantial impact that CVS can have on patients, it continues to be poorly recognized with significant delays in diagnosis and treatment, even among gastroenterologists (3). Even after diagnosis, treatment options remain limited and off-label due to the lack of US Food and Drug Administration–approved medications for CVS (6–8).

Available population-level estimates of CVS prevalence rely on patient self-reported surveys to define cases with CVS, which may not necessarily have been physician diagnosed, or include only a subset of all patients with CVS (i.e., those hospitalized for CVS) (9). Two recent studies in adults by Aziz et al and Sperber et al reported prevalence estimates of 2% in the United States, 1% in the United Kingdom, 0.7% in Canada, and 1.2% in 26 study countries (4,10). Pediatric studies in Scotland and Turkey have shown a prevalence of 1.9% in children (11,12). Existing literature indicates a need for a more in-depth epidemiological analysis that involves identification of patients with CVS based on population-level databases.

This study aimed to determine the incidence and prevalence of CVS in 2019 overall, stratified by age, sex, region, and race/ethnicity, using large US administrative claims databases composed of patients with employer-sponsored insurance and those covered under Medicaid. In addition, both clinical and treatment characteristics among patients with CVS were described.

METHODS

Study design and data source

This cross-sectional study used deidentified data from 3 Merative MarketScan US administrative claims research databases. Each provides access to health care data of patients who were insured commercially, as part of the Medicare Supplemental programs, or as part of the state Medicaid programs. The MarketScan databases include some of the largest samples available in proprietary US databases—with >263 million unique patients since 1995. MarketScan databases contain health care data for >39 million covered individuals annually, large enough to effectively estimate the incidence and prevalence of disease (13,14).

Commercial and Medicare Supplemental databases are geographically dispersed, including individuals from all 4 Census regions (Northeast, Midwest, South, and West), thus allowing for the creation of a nationally representative data sample of Americans. Both databases provide detailed health care resource use, outcomes, and cost data for medical services performed in inpatient and outpatient settings. The medical claims are linked to outpatient prescription drug claims and person-level enrollment data through unique enrollee identifiers, thus preventing duplication of records.

The MarketScan Medicaid Multi-State Database contains pooled health care data derived from millions of Medicaid enrollees located across multiple US states (though not as geographically dispersed). Regional breakdowns are not available for data reporting. The Medicaid files include records of inpatient services, inpatient admission, outpatient services, and prescription drug claims and information on race and ethnicity of enrollees.

To assess CVS estimates in 2019, MarketScan data from January 1, 2019, through December 31, 2019, were used for prevalence, and data from January 1, 2018, through December 31, 2019, were used for incidence. All variables were defined using International Classification of Diseases, 10th Revision, Clinical Modification (ICD-10-CM) codes, Current Procedural Terminology 4th edition codes, Healthcare Common Procedure Coding System codes, and National Drug Codes.

Study population and period

We used the following criteria to identify patients with CVS, their demographic and clinical characteristics, and the overall prevalence and incidence rates. Patients were required to have 1+ inpatient or 2+ outpatient claims for CVS at least 7 days apart, with a diagnosis for CVS between January 1, 2019, and December 31, 2019 (ICD-10-CM: G43.A0, G43.A1, R11.15; base case definition). For prevalence estimates, patients were required to have continuous enrollment for the entire 2019 year; for incidence rate calculations, patients were required to not only have continuous enrollment for the years 2018–2019 but also not have a CVS diagnosis before 2019 (to ensure newly diagnosed CVS in 2019).

As sensitivity analyses, 2 additional cohorts were evaluated to provide an upper bound (less-restrictive definition) and lower bound (more-restrictive definition) for incidence and prevalence of CVS. The less-restrictive definition required only 1 claim of CVS diagnosis (regardless of setting), and the more-restrictive definition required that patients met the criteria of the base case definition and have 1+ claim for nausea/vomiting in the 6 months before the earliest CVS diagnosis.

Patient sociodemographic and clinical characteristics

Patients' sociodemographic characteristics, including age, sex, payor, and setting of diagnosis (e.g., inpatient, emergency department, and other outpatient) were assessed on the date of the first CVS diagnosis. Due to limited data availability, data on geographic region was only reportable for Commercial/Medicare–insured patients, whereas race/ethnicity was only reportable for Medicaid-covered patients. The Deyo-Charlson comorbidity index and its individual component comorbidities were calculated and identified based on the data available from the 1 year before the first CVS diagnosis. Additional conditions or symptoms identified in the literature as having a higher burden in patients with CVS were also reported, including abdominal pain, anxiety, asthma, autonomic dysfunction, cannabis abuse/use, cardiac conditions (inclusive of hypertension and other heart and circulatory diseases), depression, fibromyalgia, gastroesophageal reflux disease, gastroparesis, irritable bowel syndrome, migraine, nausea, seizures (epileptic and other), smoking, and type 1 diabetes.

Treatment characteristics

Treatments for CVS were identified based on the American Neurogastroenterology and Motility Society and the Cyclic Vomiting Syndrome Association treatment recommendations and additional medications often used to treat CVS (as prophylaxis or for acute episodes) (7). The number and proportion of patients using each drug class (i.e., antidepressants, tricyclic antidepressants, atypical antidepressants/antipsychotics, and selective serotonin reuptake inhibitors, anticonvulsants, NK1 receptor antagonists, serotonin 5-HT3 receptor antagonists, antimigraine agents, and dopamine receptor antagonists) were reported separately during the 30-day period before and 90-day period after the initial CVS diagnosis. Although selective serotonin reuptake inhibitors are typically used for treating psychological comorbidity (and not specifically recommended as a treatment for CVS), they were included in the list because they are often used by clinicians in this disease population.

Incidence and prevalence

CVS prevalence rates were estimated by dividing the number of patients diagnosed with CVS in 2019 by the total population enrolled in MarketScan (Commercial, Medicare and Medicaid) in 2019. CVS incidence rates were estimated by dividing the number of patients newly diagnosed (i.e., absence of a diagnosis of CVS before 2019 using a 1-year, look back period of 2018) with CVS in 2019 by the total population in the MarketScan databases (Commercial, Medicare, and Medicaid) enrolled in 2018 and 2019. Incidence and prevalence rates were reported per 100,000 individuals for the overall cohort and stratified by age and sex subgroups (for all data sources). In addition, incidence and prevalence rates were reported by geographic region for patients with Commercial/Medicare insurance and by race/ethnicity for patients with Medicaid.

Analysis

Bivariate analyses were conducted for all study variables. Categorical variables were presented as the count and percentage of patients, whereas continuous variables were summarized by providing the mean and SD. Patient sociodemographic, clinical, and treatment characteristics were reported for all newly diagnosed patients with CVS, divided into pediatric (younger than 18 years) and adult cohorts (18 years or older).

RESULTS

Study population

The total population with enrollment in the full year 2019 in MarketScan Commercial or Medicare was 15,045,070 (for prevalence estimates) and those with enrollment in both 2018 and 2019 was 11,872,021 (for incidence estimates). Of these, we identified a total of 2,513 prevalent patients with CVS and 1,263 patients newly diagnosed (incident) with CVS in 2019. These patients all met eligibility criteria using the base case definition among those with employer-sponsored insurance (Commercial or Medicare). Among those enrolled in Medicaid (total with enrollment in full year 2019 was 7,888,771 [denominator for prevalence estimates], total with enrollment in both 2018 and 2019 was 5,669,915 [denominator for the incidence estimates]), there were 3,382 and 1,507 patients with prevalent and newly diagnosed (incident) CVS using the base case definition (Figure 1), respectively. Patients identified as having prevalent or incident CVS using either the upper bound (less restrictive) or lower bound (more restrictive) case definitions are included in Figure 1.

Figure 1.

Patients with prevalent and incident CVS in 2019 MarketScan commercial, Medicare supplemental, and Medicaid databases. CVS, cyclic vomiting syndrome; ICD-10-CM, International Classification of Diseases, 10th Revision, Clinical Modification.

Prevalence and incidence rates

The prevalence rate of CVS was 16.7 (Commercial/Medicare) and 42.9 (Medicaid) per 100,000 individuals in 2019. Subgroup analyses found higher prevalence rates in female individuals (female individuals: 19.9, 48.4 per 100,000 individuals; male individuals: 13.3, 36.0 per 100,000 individuals) in Commercial/Medicare and Medicaid, respectively. Individuals with employer-sponsored insurance (Commercial/Medicare) had higher prevalence in those aged 5–13, 14–17, and 18–24 years (Table 1, Figures 2 and 3), whereas the Medicaid-insured population had peak prevalence rates in those aged 18–24, 24–44, and 45–64 years (Table 1, Figures 2 and 3). The prevalence of CVS across geographic regions within the United States (available only in Commercial/Medicare data), ranged from 14.2 per 100,000 in the Northeast to 19.3 in the Midwest (Table 1, Figure 4). In addition, we observed differences in the prevalence of CVS across racial categories (available only in Medicaid data): White/Caucasian, 48.6 per 100,000; Black/African American, 35.3 per 100,000; and Hispanic, 25.5 per 100,000 (Table 1, Figure 4). The observed incidence rate of CVS in 2019 was 10.6 (Commercial/Medicare) and 26.6 (Medicaid) per 100,000 individuals. We observed similar trends in the variation of incidence rates across sociodemographic factors (age, sex, US region, and race/ethnicity) (Table 1, Figures 2–4).

Table 1.

Prevalence and incidence of CVS in 2019: stratified by age, sex, and region

Figure 2.

Overall incidence and prevalence of CVS in 2019 (per 100,000 person-years). (a) Incidence. (b) Prevalence. *Base case CVS definition: at least 1 inpatient or 2 outpatient claims for CVS at least 7 days apart. **Sensitivity 1 less restrictive: at least 1 inpatient or outpatient claim for CVS. ***Sensitivity 2 more restrictive: at least 1 inpatient or 2 outpatient claims for CVS at least 7 days apart and at least 1 claim for nausea/vomiting in the 6 months preceding diagnosis (including diagnosis date). CVS, cyclic vomiting syndrome.

Figure 3.

Incidence and prevalence rates of CVS among Commercial/Medicare and Medicaid US populations in 2019 by age groups. (a) Incidence. (b) Prevalence. CVS, cyclic vomiting syndrome.

Figure 4.

Incidence and prevalence rates of CVS in 2019 stratified by region or race. (a) Region. (b) Race/ethnicity. CVS, cyclic vomiting syndrome.

In addition to the base case, we conducted sensitivity analyses to set bounds in the estimates of CVS epidemiology. The prevalence estimates using a less-restrictive case definition were higher (53.1 and 132.8 per 100,000 in Commercial/Medicare and Medicaid insured, respectively), while those based on more-restrictive case definition were lower (11.4 and 33.0 per 100,000). The sensitivity analyses for incidence estimates demonstrated similar trends (Figure 2).

Sociodemographic characteristics

Among 1,263 patients defined as incident CVS with Commercial/Medicare supplemental insurance, the mean age was 33.1 (SD 20.5) years, 62.5% were female, and 94.9% had Commercial insurance (5.1% Medicare Supplemental). Most (72.1%) of the cases with incident CVS were diagnosed as adults, with more than half of all incident cases derived from those aged 25–44 (22.6%) and 45–64 (26.3%) years. Among the 1,507 patients with incident CVS covered by Medicaid, the mean age was slightly younger at 28.1 (18.0) years with 64.2% of these cases diagnosed as adults. Additional demographic results showing regional (for Commercial/Medicare) and race/ethnicity (for Medicaid) distributions of incident CVS cases are summarized in Table 2.

Table 2.

Demographic characteristics of patients with incident CVS measured on date of diagnosis

Clinical characteristics

Among all incident patients with CVS, 37.2% (Commercial/Medicare Supplemental) and 42.3% (Medicaid) received their first diagnosis during an inpatient admission (with a larger proportion in the adult cohort [45%–55%] vs that in the pediatric cohort [16%–20%]) (Table 2). The mean Deyo-Charlson comorbidity index score was 1.1 (2.1) and 1.5 (2.1) among all patients with CVS in Commercial/Medicare and Medicaid, respectively. The most common CVS-related comorbid conditions or symptoms were abdominal pain, cardiac conditions, and anxiety. Results were consistent among Commercial/Medicare and Medicaid patients and are summarized in Table 3.

Table 3.

Clinical characteristics of patients with incident CVS

We found that a notable proportion of patients (21.3% of Commercial/Medicare patients and 28.1% of Medicaid patients) were prescribed medications that could function as prophylaxis for CVS in the 30 days before the first CVS diagnosis (Table 4). Similarly, 34.9% of Commercial/Medicare patients and 39.3% of Medicaid patients were prescribed medications that could be used as acute, abortive treatment for CVS symptoms in the 30 days before the first CVS diagnosis (Table 4). Only 32%–35% had prescriptions for prophylactic treatment and 47%–55% for acute treatment within the first 30-day period following their diagnosis. The prevalence of prescriptions for prophylaxis treatments increased modestly in the 30-day period after CVS diagnosis (Commercial/Medicare: 10.8% increase; Medicaid: 6.8% increase) and continued to increase through the 90-day period after initial diagnosis (Commercial/Medicare: 22.2% increase; Medicaid: 15.8% increase) (Table 4). Similarly, prescriptions for acute treatments increased in the 30-day and 90-day periods following CVS diagnosis (Table 4).

Table 4.

CVS-related treatment among patients with incident CVS

DISCUSSION

This study provides the first comprehensive estimates of CVS prevalence and incidence rates using large-sized US administrative claims databases, capturing patients in all health care settings and covered by 3 major insurance types. To mitigate bias due to the COVID-19 pandemic (15), we selected the 2019 data for better quality and more meaningful interpretation. In this study, the estimated prevalence rates of CVS were 16.7 and 42.9 and the incidence rates were 10.6 and 26.6 per 100,000 individuals with employer-sponsored insurance and Medicaid coverage, respectively. Given the lack of disease awareness and underdiagnosis of CVS in the medical community, these estimates are conservative, and the true incidence and prevalence of CVS is likely greater than what we report here. In addition, cases with CVS defined by disease-specific ICD-10 codes and reported by providers are more reliable for epidemiological estimation, compared with prior research using surveys based on patient self-report symptoms with prevalence rates of 1%–2% (i.e., approximately 1,000–2,000 per 100,000) (4,10–12). The only prior study using population-level data in the United States assessed a subset of patients with a hospitalization for CVS from 2005 to 2014; the study focused on cannabis use outcomes, and cases with CVS were identified with ICD-9 codes, which are less specific (including a broader list of conditions such as gastroparesis, chronic nausea and vomiting etc) (9).

In this study, CVS incidence and prevalence rates per 100,000 patients (Figure 3a, b) were notably higher among individuals covered by Medicaid than those with employer-sponsored insurance in the non-older adult groups (i.e., aged 18–64 years). It is not surprising to observe higher CVS rates among those with Medicaid coverage because it is a population with socioeconomic factors susceptible to increased prevalence of chronic disease compared with the general United States (and even with other low-income populations) (16,17). One potential explanation for the higher rates in the adult age groups (vs pediatric age groups) relates to access-to-care barriers among the Medicaid-insured populations. Historically, Medicaid populations are less likely to receive preventive care such as screening and timely referrals to specialty care, regardless of disease areas (18–20). Higher incidence and prevalence rates in the non-older Medicaid populations may relate to various factors, for example, lack of regular care resulting in delayed diagnosis from pediatric to adult ages or becoming Medicaid eligible due to employment loss and/or disability caused by disease. Essentially, the CVS incidence and prevalence rates in the Medicaid-insured population that we reported in this study are likely underestimated. Using less restrictive criteria for CVS diagnosis, the incidence and prevalence were 105.7 and 132.8 per 100,000.

Abdominal pain, cardiac conditions, and anxiety were the most common comorbid conditions in our cohort of patients with CVS, and prevalence was higher among adults (vs children). Abdominal pain and anxiety were reported in 59% and 32% of adults with CVS, respectively, which was notably higher compared with general population estimates (e.g., 12-month prevalence rates of anxiety range from 4% to 19% [increasing in recent years] and abdominal pain 22% to 25%) (21–24). This study found that 51% of adults with CVS had a cardiac condition, which is consistent with the updated annual report from the American Heart Association and a study using administrative claims data (e.g., both studies reported that approximately 49% of the general population had cardiovascular conditions) (25,26). Our analysis also captured prescriptions dispensed potentially for prophylaxis or acute treatment among patients with CVS during the periods immediately preceding and following CVS diagnosis. Although treatment rates increased after diagnosis, only 32%–35% had a prescription for prophylaxis treatment and 47%–55% for acute treatment within the first 30 days. In addition, the fact that 45%–55% of the adult cohort received their CVS diagnosis in the inpatient setting (vs 16%–20% of the pediatric cohort) may indicate a lower level of CVS awareness in the outpatient setting among both gastroenterologists and primary care physicians for adults and poorer CVS disease control. These data indicate that there is an unmet need for additional resources for physician/provider education to raise awareness about CVS and its management.

The primary strength of this analysis was the use of large claims databases with a physician diagnosis of CVS as opposed to patient reports that can be inaccurate. In addition, these data represent a national footprint in the United States: the MarketScan Commercial and Medicare Supplemental databases include data from individuals in all US Census regions; the Medicaid claims data include patient-level race/ethnicity information, which offers an important aspect to better understand health care disparities among racial/ethnic groups. Second, although prevalence and incidence rates of CVS using administrative claims are conservative estimates for this underdiagnosed condition, the aggregation of cases across all medical service settings in the US health care delivery system offered a more comprehensive view of disease epidemiology than what was previously reported. In addition, continuous enrollment during the study years enabled longitudinal tracking of individuals, thus minimizing attrition of the study population, and preventing overestimates of prevalence and incident rates. Last, the exploration of multiple versions of the CVS case definition (i.e., base definition and 2 sensitivity analyses definitions) provides a range for estimation.

Limitations of this study include those inherent in any analysis using real-world administrative claims data, such as lack of clinical detail to understand the reason for events (e.g., because all medications are prescribed off-label for CVS treatment, it is unknown whether the prescription purpose was for CVS management or for other comorbid conditions such as anxiety, depression, or migraine) and the possibility of data bias due to coding practices (e.g., different medical professionals may have different coding patterns and coding may be inaccurate when assignment of the billing code is not done by the clinician). All administrative claims data rely on medical codes to identify conditions that may be subject to misclassification; however, ICD-10 has specific codes for CVS, which may mitigate this error. Additional work is needed in the field to further explore comorbidities found in association with CVS. This study was limited to individuals with either employer-sponsored insurance coverage or with Medicaid; and consequently, results may not be generalizable to patients who are covered by other insurance types or those who are uninsured/underinsured. Last, because administrative claims databases may not capture patients' full health history, this analysis included a 1-year washout period to define incident patients. This approach may still not have removed all previously diagnosed patients (i.e., a patient could be included as an incident case in 2019 if they received a CVS diagnosis earlier than 2018 or was covered by other insurance types). Thus, it is possible that our incidence estimates could have been inflated.

In summary, this analysis provides the first estimates of incidence and prevalence rates of CVS using large claims databases across all health care settings and covering 3 major insurance types in the United States. Our findings also showed the patterns of medication use in the time frame context of CVS diagnosis. These data may help inform strategies to address challenges and unmet needs associated with diagnosis and treatment of CVS. Specifically, these findings underscore the need for better recognition and prompt management of patients with CVS. Education of the medical community and the development of better screening strategies and more effective treatment options are warranted.

CONFLICTS OF INTEREST

Guarantor of the article: Camilla Richmond, MD.

Specific author contributions: Y.J.C., N.P., and I.W.: contributed to conception, design, and planning of the study, analysis of the data, interpretation of the results, drafting of the manuscript, and approving the final draft submitted. C.R., J.W., A.T., T.V., and D.J.L.: contributed to conception of the study and interpretation of the results, critically reviewed the manuscript for important intellectual content, and approved the final draft submitted.

Financial support: This study was funded by Takeda Development Center Americas.

Potential competing interests: Y.J.C., C.R., and J.W. are current or former employees of Takeda Development Center Americas, Inc., and own stock or stock options. N.P. and I.W. are/were employed by Merative, which received funding from Takeda to conduct this study. D.J.L. is a consultant for Takeda, Alexza, and Aditum Pharmaceuticals. T.V. is a consultant for Takeda. A.T. has nothing to disclose.

Study Highlights.

WHAT IS KNOWN

• ✓ There is a scarcity of population-level and subpopulation-level epidemiological information on individuals diagnosed with cyclic vomiting syndrome (CVS).

• ✓ CVS imposes a substantial burden on patients, partly due to poor clinical recognition and significant delays in diagnosis and treatment.

WHAT IS NEW HERE

• ✓ This study provides the first population-level estimates of CVS incidence and prevalence using all service-setting claims data from 3 major insurance types in the United States.

• ✓ The CVS prevalence and incidence in 2019 were 16.7 and 10.6 per 100,000 individuals covered by Commercial/Medicare Supplemental insurance, and 42.9 and 26.6 per 100,000 individuals covered by Medicaid, respectively.

• ✓ Despite a diagnosis of CVS, only 32%–35% had a prescription for treatment used for prophylaxis and 47%–55% had a prescription for acute treatment within the first 30 days following initial diagnosis.

• ✓ These findings are important for improving disease awareness, which may lead to the development of better screening strategies and prompt and effective treatments.