1 INTRODUCTION

Cyclic vomiting syndrome (CVS) is an idiopathic disorder characterized by recurrent, stereotypical bouts of intense nausea and vomiting that are often accompanied by abdominal pain.1 The pathophysiology of CVS remains poorly understood, but it is generally believed to reflect dysregulation of central nervous system control over autonomic function and emesis.2-4 Some clinical features of CVS, including the observation that CVS attacks can often be “triggered” by intense stressors or emotions, highlight the major role that mind-body interactions play in this disorder.3 Furthermore, CVS attacks have an unpredictable nature, often occurring “out of the blue”. CVS patients often remark that this “unknown” aspect—that is, not knowing when the next CVS attack will come—is a significant source of anticipatory anxiety. Adults with CVS have high rates of comorbid psychiatric disorders, particularly anxiety-spectrum diagnoses such a generalized anxiety or panic disorder.5-7 Thus, a substantial proportion of CVS patients may be particularly vulnerable to anticipatory anxiety due to a lack of robust cognitive and emotional coping mechanisms.8-10 Clearly, the physical burden of frequent or prolonged CVS attacks and the psychological impact of living with an unpredictable disorder both could contribute to poorer QOL in adult CVS patients.

Recent clinical guidelines for the diagnosis and management of CVS in adults emphasize the important role that comorbid psychiatric disorders, particularly anxiety-spectrum illnesses, may have in the expression of CVS and in mediating negative impacts on patient QOL.11 Indeed, these guidelines recommend that patients diagnosed with CVS should not only be screened for comorbid mood disorders, but that complementary or psychological therapies designed to reduce anxiety and blunt responses to stress should form a component of comprehensive CVS management.11, 12 However, it is unclear which patients with CVS may particularly benefit from these more comprehensive, often resource-intensive therapies.

Intolerance to Uncertainty (IU) is a cognitive vulnerability that predicts feeling distressed during unpredictable situations, particularly those perceived as having potentially negative consequences.13 The IU concept incorporates emotional, cognitive, and behavioral responses when faced with uncertain situations.14 Higher trait IU is viewed as a cognitive vulnerability factor that is strongly associated with higher lifetime risk of developing multiple distinct psychiatric conditions, particularly those associated with aberrant levels of fear and worry (eg, general anxiety disorder, panic disorder, obsessive-compulsive disorder).13-15 Although some limited evidence suggests that higher IU may also associate with poorer QOL in some medical conditions such as multiple sclerosis or cancer,16-18 the potential influence of higher IU on QOL in disorders of gut-brain interaction (DGBIs) remains unexplored. We hypothesized that patients with higher IU would have worsened quality of life, more frequent CVS attacks, and overall higher healthcare utilization compared to those with lower IU. To test our hypothesis, we conducted a cross-sectional survey of adults diagnosed with CVS and obtained demographic information, clinical features of their CVS disorder, and standardized assessments of IU, mood disorders, and QOL.

2 MATERIALS AND METHODS

We collected data from adults (aged 18 or older) previously diagnosed with cyclic vomiting syndrome by an expert CVS clinician (Dr. Thangam Venkatesan) at the Medical College of Wisconsin. Subjects who met Rome IV criteria for CVS and who had been previously enrolled in a clinical patient registry in REDCap (Research Electronic Database Capture) formed the pool of eligible participants. After appropriate approval by the MCW Institutional Review Board (IRB) in June 2017 (MCW IRB # PRO29813), patients in the registry were contacted by email with a brief description of the proposed study and an invitation to participate. As an incentive to participate in the study survey, all subjects were entered into a raffle to win a new iPad tablet (Apple Inc.; retail value ~$500). Those interested were then directed to a weblink with the consent form. After informed consent was obtained, subjects completed multiple questionnaires which assessed: 1) clinical features of CVS, comorbid conditions (such as anxiety and depression), and substance use, 2) patterns of healthcare utilization including emergency department (ED) visits, hospitalizations, and telephone calls to clinics, and 3) validated assessments of psychosocial function and quality of life. We used REDCap to build and administer the online surveys and to collect data. This web application allows subjects to enter only a single survey from a single IP address. All data that were subsequently analyzed from REDCap were viewed in a de-identified format by coinvestigators from the University of Pittsburgh. Some demographic and clinical information from study entrants had previously been recorded into the REDCap patient registry, and this information was incorporated into the final de-identified dataset.

2.1 Study participants

The MCW registry included 653 adult patients diagnosed with CVS who had available email contact information. All of these individuals were sent an email inviting participation. 122 engaged with the study to provide consent and begin some participation, but ultimately 118 individuals completed both the consent process and the full study questionnaire between August and December 2017. Thus, the final participation rate was ~18%.

2.2 Demographic and clinical information

Subjects were specifically asked to list their age (years), gender, racial background, and education level obtained, and patterns of substance use (marijuana, cocaine, or heroin use within last 6 months, cigarette/e-cigarette use and amount, regular alcohol use and amount). We asked subjects about the presence of various medical conditions, including mood disorders. Subjects also provided information about their CVS experience, including duration of CVS symptoms, the presence of any circadian pattern to their CVS attacks (defined as >50% of attacks tending to occur in any specific quartile of the day referenced to a midnight to midnight cycle), the presence and length (minutes) of any prodrome preceding CVS attacks, the length of CVS episodes as defined by active vomiting (hours), typical number of vomiting episodes / hour during an attack, and the typical length of the recovery period (hours after last vomit).

2.3 Validated study instruments

We administered the Intolerance to Uncertainty Scale—Short Form (IUS-12; Supplemental Table S3),19 PROMIS Scale v1.1—Global Health survey,20 and the State-Trait Anxiety Inventory (STAI).21 Possible IU scores range from 12 to 60, with higher values indicating greater degrees of intolerance to uncertainty. The mean IU score for the entire cohort was 29.4 ± 9.9. For the purposes of dichotomizing the study group into those with higher IU and lower IU, we used a cutoff IU score of 35 or higher to define higher IU.15

2.4 CVS Prodromal panic attack

Subjects were asked whether intense fear or discomfort and any of 13 additional symptoms were present during the prodromal phase of their CVS attacks. If 4 or more of these 13 symptoms were present (palpitations, pounding heart, or accelerated heart rate, sweating, trembling or shaking, sensations of shortness of breath or smothering, feeling of choking, chest pain or discomfort, nausea or abdominal distress, feeling dizzy, unsteady, lightheaded, or faint, derealization [feelings of unreality] or depersonalization [being detached from oneself], fear of losing control or “going crazy”, fear of dying, paresthesia [numbness or tingling sensation], or chills or hot flushes), then the subject was deemed to meet DSM-IV criteria for a panic attack that occurred during the CVS prodrome phase.22

2.5 Statistical analysis

Characteristics of the study cohort are presented as n (%) for categorical variables and median (range) for continuous variables. For many of the study analyses, we divided patients into higher IU (score of 35 or higher) versus lower IU (score of 34 or lower) groups. We performed group comparisons (higher IU versus lower IU) on the demographic data, clinical parameters, the PROMIS Physical and Mental subscales, and the STAI scores using Fisher's exact tests for categorical variables and Mann-Whitney tests for continuous variables. We present scatterplots to visually represent the association between IU and the respective PROMIS and STAI measures and boxplots to show the distribution of each measure for higher IU versus lower IU patients. We present Pearson correlation coefficients for the relationship between IU and the respective PROMIS and STAI measures and other health-related outcomes (eg, urgent care visits, hospitalizations, episodes of nausea, and vomiting). Finally, we performed logistic regression analyses to test whether increasing IU score was associated with higher odds of negative quality of life outcomes (eg, loss of job, applying for disability, delaying education). We initially performed univariate analyses and then adjusted for possible confounding due to age and sex covariates. p values <0.05 were considered significant.

2.6 Missing data

Only a few demographic and clinical data points were omitted by study participants. For example, age and weight information were not available for some participants, and smoking status was not entered by others. When presented, the range (median) is derived from available data, and the proportional (%) data uses the number of non-missing values as the denominator for the measure of interest. Missing data also modestly impacted the scoring and inclusion of the standardized questionnaires. By convention, if 1–2 questions are left unanswered in the STAI, then the mean of the remaining items are multiplied by 20 and rounded up to the higher whole number to derive an estimated score. This procedure was used in only a few instances. The PROMIS questionnaires do require that all items are completed to generate a valid score. There were two subjects with missing data such that one subject had a missing PROMIS physical component score and another subject had a missing PROMIS mental component score.

3 RESULTS 3.1 Demographics and clinical characteristics of patients with CVS

There were 118 patients with CVS who completed the study. There were 87 (74%) patients in the group with lower IU scores (34 or lower) and 31 (26%) patients with higher IU scores (35 or higher). The demographic and clinical information of these two groups, and for the total cohort, is shown in Table 1. In general, demographic variables were similar between groups. However, there were notable differences with significantly higher rates of anxiety (93% vs. 58%, p = 0.001) and depression (76% vs. 41%, p = 0.002) in the higher IU group compared with the lower IU group. The rates of other comorbidities, such as migraines, fibromyalgia, chronic fatigue syndrome, and irritable bowel syndrome did not significantly differ between the two groups.

TABLE 1. Demographic and clinical features of the study cohort Category Category Low IU (n = 87) High IU (n = 31) Total (n = 118) p-value Demographics

Age (years)

Median (range)

Years 34 (18–70) 39 (19–59) 37 (18–70) 0.68 Gender Female 64 (73.6%) 27 (87.1%) 91 (77.1%) 0.143 Male 23 (26.4%) 4 (12.9%) 27 (22.9%) Race White 72 (82.8%) 26 (83.9%) 98 (83.1%) 1 African American 7 (8.0%) 2 (6.5%) 9 (7.6%) Other 8 (9.2%) 3 (9.7%) 11 (9.3%) Body Mass Index (BMI) 26 (16.4–74) 33.5 (24–43) 28.5 (16.4–74) 0.047 Education High school 11 (12.6%) 1 (3.2%) 12 (10.2%) 0.145 Some college 22 (25.3%) 8 (25.8%) 30 (25.4%) Associates 8 (9.2%) 8 (25.8%) 16 (13.6%) Bachelors 28 (32.2%) 10 (32.3%) 38 (32.2%) Graduate degree 18 (20.7%) 4 (12.9%) 22 (18.6%) Substance use Current smoker cigarettes/e-Cig (nicotine) Yes 10 (11.5%) 8 (25.8%) 18 (15.3%) 0.079 Ever smoker Yes 28 (35%) 16 (53.3%) 44 (40%) 0.087 Regular alcohol use Yes 16 (18.4%) 4 (12.9%) 20 (16.9%) 0.586 Marijuana use (last 6 months) Yes 22 (25.3%) 10 (32.3%) 32 (27.1%) 0.479 Illicit drugs (Heroin, Cocaine, etc.) Yes 2 (2.3% 0 (0%) 2 (1.7%) 1 Comorbid conditions Medical IBS Yes 19 (23.5%) 13 (43.3%) 32 (28.8%) 0.058 Chronic fatigue syndrome Yes 13 (16%) 2 (6.9%) 15 (13.6%) 0.346 Fibromyalgia Yes 8 (9.9%) 5 (16.7%) 13 (11.7%) 0.332 Migraines Yes 45 (57%) 21 (72.4%) 66 (61.1%) 0.183 History of cholecystectomy Yes 22 (27.2%) 10 (34.5%) 32 (29.1%) 0.481 Mood disorders Anxiety Yes 47 (58%) 27 (93.1%) 74 (67.3%) <0.001 Depression Yes 33 (40.7%) 22 (75.9%) 55 (50%) 0.002 Bipolar disorder Yes 4 (4.9%) 5 (16.7%) 9 (8.1%) 0.058 p values <0.05 are shown in bold. 3.2 Characteristics of CVS episodes and health care utilization

The features of CVS illness in the study cohort are presented in Table 2. In general, the findings confirm that a majority (61%) of CVS patients experience a prodrome that, if present, tends to be short-lived (ie, <60 min in most cases). Nearly half of those with a prodrome experienced symptoms meeting criteria as a prodromal panic attack. Attacks also tended to have a circadian pattern, with over 50% of subjects reporting an attack onset in the morning hours (between midnight and noon). The duration of a CVS attack in our cohort ranged widely from <1 h to as long as 14 days. The majority (76%) of patients found sleep to be helpful in alleviating symptoms. Nearly one quarter of the cohort tried cannabis to abort a CVS episode, with the vast majority (~95%) of those that used cannabis in this manner reporting that doing so was helpful during an episode. More than half the cohort (52%) reported hot water-bathing during an episode and of these patients, 75% reported relief with this behavior.

TABLE 2. Features of cyclic vomiting syndrome in the study cohort Features of CVS Illness Category Low IU (n = 87) High IU (n = 31) Total (n = 118) p-value Duration with CVS Years 10 (1–68) 10 (1–27) 10 (1–68) 1 Duration suffered prior to CVS diagnosis Years 4 (1–68) 5 (1–20) 4 (1–68) 0.699 CVS prodrome symptoms Yes 53 (60.9%) 19 (63.3%) 72 (61.0%) 0.971 Prodromal panic attacks Yes 19 (21.8) 14 (45.2%) 33 (28%) 0.019 CVS prodrome length Minutes 50 (1–10,080) 60 (1–1440) 55 (1–10,080) 0.610 CVS attack length Hours 12 (0.5–336) 6 (2–120) 12 (0.5–336) 0.432 Recovery period Hours 14 (1–216) 24 (1–168) 16 (1–216) 0.574 Total # CVS attacks in past year 0 10 (11.5%) 3 (9.7%) 13 (11%) 0.747 1–3 15 (17.2%) 3 (9.7%) 18 (15.3%) 4–9 18 (20.7%) 8 (25.8%) 26 (22%) 10–14 13 (14.9%) 3 (9.7%) 16 (13.6%) 15+ 31 (35.6%) 14 (45.2%) 45 (38.1%) # emeses / hour during worst episode 10 (3–50) 10 (3–100) 10 (3–100) 0.731 Onset time of >50% of CVS attacks 12 AM−6 AM 26 (29.9%) 14 (45.2%) 40 (33.9%) 0.374 6 AM−12 PM 19 (21.8%) 5 (16.1%) 24 (20.3%) 12 PM−6 PM 4 (4.6%) 1 (3.2%) 5 (4.2%) 6 PM−12 AM 12 (13.8%) 6 (19.4%) 18 (15.3%) No pattern 26 (29.9%) 5 (16.1%) 31 (26.3%) Take hot showers/baths during attack Yes 41 (51.2%) 16 (55.2%) 57 (52.3%) 0.829 -if yes, alleviates symptoms Yes 32 (78%) 11 (68.8%) 43 (75.4%) 0.505 Use Marijuana during attack Yes 17 (22.7%) 7 (25.9%) 24 (23.5%) 0.793 -if yes, alleviates symptoms Yes 16 (94.1%) 6 (85.7%) 22 (91.7%) 1 Does sleep help symptoms? Yes 59 (74.7%) 22 (78.6%) 81 (75.7%) 0.801 Symptom Free between CVS Attacks Yes 56 (70.9%) 18 (62.1%) 74 (68.5%) 0.484 Healthcare utilization in the last year (related to CVS) # Urgent care / ED visits (not admitted) 0 42 (48.3%) 16 (51.6%) 58 (49.2%) 0.735 1–3 26 (29.9%) 7 (22.6%) 33 (28%) 4+ 19 (21.8%) 8 (25.8%) 27 (22.9%) # Overnight hospitalizations 0 69 (79.3%) 26 (83.9%) 95 (80.5%) 0.912 1–3 14 (16.1%) 4 (12.9%) 18 (15.3%) 4+ 4 (4.6%) 1 (3.2%) 5 (4.2%) # calls/emails to provider 0 44 (50.6%) 19 (61.3%) 63 (53.4%) 0.57 1–3 19 (21.8%) 6 (19.4%) 25 (21.2%) 4+ 24 (27.6%) 6 (19.4%) 30 (25.4%) Impact of CVS disorder on work/life Problems with job resulting from CVS Yes 52 (65.8%) 24 (88.9%) 76 (71.7%) 0.026 Lost Job due to CVS Yes 19 (24.7%) 11 (40.7%) 30 (28.8%) 0.14 Applied for disability due to CVS Yes 18 (24%) 9 (34.6%) 27 (26.7%) 0.312 Awarded disability benefits Yes 8 (11.4%) 7 (29.2%) 15 (16%) 0.055 Delayed higher education due to CVS Yes 28 (38.4%) 10 (37%) 38 (38%) 1 p values <0.05 are shown in bold.

Approximately half (51.7%) of the cohort experienced 10 or more attacks in the previous year, and not surprisingly, CVS had a major impact on healthcare utilization. Nearly half of the cohort reported visiting urgent care / emergency departments due to CVS in the past year at least once, with a subset (19%) of patients requiring hospitalization. Finally, a CVS diagnosis was associated with a substantial impact on work and life planning. The vast majority (71%) of the cohort reported problems with work and a substantial proportion (29%) of individuals lost a job due to CVS. About a quarter (27%) of the cohort had applied for disability benefits, with about one in six CVS patients reporting being awarded disability due to their illness. More than a third (38%) reported delays in higher education due to their CVS symptoms.

We compared the reported features of CVS illness in those with higher IU compared with lower IU scores, as shown in Table 2. There were no significant differences between groups observed in several features of the illness, such as the duration and frequency of CVS attacks, the presence of prodromal symptoms, the circadian pattern of attacks, the presence of factors that alleviated CVS attacks, or patterns of healthcare utilization. However, a greater proportion of higher IU scorers reported prodromal panic attacks (45% vs. 22%, p = 0.01) and problems with their job due to CVS (89% vs. 67%, p = 0.02) compared with lower IU scorers.

3.3 Correlation between IU scores with quality of life and state/trait anxiety

We then investigated the associations between IU scores with standardized measures of anxiety and both mental and physical aspects of quality of life (Supplemental Table S1). We observed group differences in each of these measures with higher IU scorers reporting worse mental and physical quality of life, and more severe state and trait anxiety, compared to those with lower IU scores (Table 3). IU scores most strongly correlated negatively with mental aspects of quality of life (Figure 1A), with higher IU scorers reporting a significantly lower mean PROMIS mental score than lower IU scorers (Figure 1B). IU scores also correlated negatively with physical aspects of quality of life (Figure 2A), with higher IU scorers reporting a significantly lower mean PROMIS physical score than lower IU scorers (Figure 2B). The lower mean PROMIS mental and physical scores reported by high IU scorers are more than 1 standard deviation below the PROMIS scores of a reference, healthy population, consistent with a significant decrement in quality of life in this group. IU scores in our cohort most strongly correlated with trait anxiety as measured by the STAI (Figure 3A), with higher IU scorers reporting a significantly higher mean STAI trait score than lower IU scorers (Figure 3B). The higher mean STAI scores reported by high IU scorers are consistent with a clinically significant increase in anxiety symptoms, compared with a reference, healthy population.

TABLE 3. Impact of IU on measures of anxiety and quality of life Low IU (n = 87) High IU (n = 31) Total (n = 118) Unadjusted p value Adjusteda p value STAI State Total 37.1 ± 10.6 55.2 ± 9.4 41.8 ± 13.0 p < 0.001 p < 0.001 STAI Trait Total 39.2 ± 9.9 57.4 ± 7.9 44.0 ± 12.3 p < 0.001 p < 0.001 PROMIS Physical T-Score 45.7 ± 8.8 37.0 ± 7.4 43.4 ± 9.3 p < 0.001 p < 0.001 PROMIS Mental T-Score 45.7 ± 9.0 34.6 ± 6.2 42.8 ± 9.7 p < 0.001 p < 0.001 a Adjusted for age and gender.