Increased Risk of Hospitalization in Celiac Disease With COVID-19 Infection Is Mitigated by Vaccination

Abstract

Background

We sought to describe clinical characteristics of celiac disease (CD) patients infected with coronavirus disease 2019 (COVID-19) and estimate hospitalization risk, intensive care unit (ICU) requirement, mortality, and thrombosis, and the impact of vaccination on these outcomes.

Methods

We performed a single-center, retrospective cohort study comparing biopsy-proven CD patients with a matched sample of non-CD (referent) patients diagnosed with COVID-19 between March 2020 and January 2022. Matching ensured 2 referent patients for every 1 CD patient by age, sex, ethnicity, and COVID-19 diagnosis date. We also adjusted for general and celiac-specific comorbidity. The primary outcome was hospitalization. Secondary outcomes included ICU requirement, mortality, and thrombosis. We also compared these outcomes between vaccinated and unvaccinated individuals.

Results

We included 330 patients: 110 with CD (mean age 47 years, 83% female) and 220 matched referents. Hospitalization occurred in 27 CD patients (24%) and 25 referent patients (11%) (hazard ratio, 2.10; 95% confidence interval, 1.21–3.65; P = .009). Vaccination was associated with significantly decreased risk of hospitalization (hazard ratio, 0.53; 95% confidence interval, 0.31–0.93; P = .026). Four unvaccinated CD patients and 2 unvaccinated referent patients required ICU. No mortality occurred among CD patients, and 2 referent patients died. No thrombosis occurred in either group.

Conclusions

CD patients with COVID-19 have a higher risk of hospitalization compared with non-CD referents. This risk is mitigated by vaccination in CD patients as it is in non-CD referents. ICU requirement occurred only in unvaccinated CD patients, and no CD patient died. Vaccination against COVID-19 should be strongly recommended in patients with CD as it is for non-CD patients in the general population.

Graphical abstract

What You Need to Know.

Background

Celiac disease is associated with higher susceptibility to community-acquired respiratory illnesses.

Vaccination against COVID-19 is an effective means of protection.

Findings

Celiac disease appears to convey a higher risk for hospitalization with COVID-19 infection.

Vaccination against COVID-19 is effective in mitigating hospitalization risk in celiac disease patients.

Implications for patient care

Vaccination against COVID-19 should be strongly recommended in patients with celiac disease.

Severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) has infected millions of individuals globally, with upward of 6.6 million fatalities attributable to the coronavirus disease 2019 (COVID-19) as of January 2023 according the World Health Organization.¹ Several risk factors have been found to be associated with severe disease, hospitalization, and mortality such as advanced age, immunosuppression, and certain comorbidities including diabetes mellitus, advanced liver disease, and preexisting lung disease.2, 3, 4, 5

Celiac disease (CD) is a multisystem condition with a global prevalence of approximately 1%.6, 7, 8, 9 Although studies have shown increased susceptibility to viral illnesses, the research to date has demonstrated similar incidences and outcomes between those with CD with COVID-19 and the general population.8, 10, 11, 12, 13 No mortality was documented in an observational study including 11 patients with treated CD from Iran.¹⁴ Although the humoral response to SARS-CoV-2 vaccines was similar in CD to that observed in healthy controls,¹⁵ prior reports about outcomes of CD and COVID-19 were dated before the availability of COVID-19 vaccines. Therefore, the effect of vaccination on COVID-19 outcomes in patients with CD is unknown. Because new strains of the virus such as the Omicron and BA.2 variants led to continued surges, it is important to understand related outcomes in CD patients at different stages of the epidemic.

In this study, we aimed to compare the frequency of COVID-19–related hospitalization between patients with and without CD before and after vaccination. We also compared these patients with regard to the rates of intensive care unit (ICU) requirement, mortality, and thrombosis. We hypothesized that although hospitalization rates would be similar between those with and without CD, there would be improvement in this outcome and others among those who had been vaccinated.

Methods

We performed a retrospective cohort study at Cleveland Clinic (a tertiary care center) using data collected from patients’ electronic medical records. The study was approved by our institutional review board.

We obtained data from the electronic medical records system by screening for all adult (age >18 years) patients with CD who tested positive for SARS-Cov-2 infection in any of the clinic laboratories between March 1, 2020 and January 1, 2022. COVID-19 diagnosis was defined as a positive test using nucleic acid amplification test real-time polymerase chain reaction in respiratory tract specimens as recommended by Centers for Disease Control and Prevention. We then matched these patients on the basis of age (±1 year), ethnicity, sex, and date of COVID-19 diagnosis (±3 months) with a referent group consisting of adults without a clinical diagnosis of CD as documented in the medical record who had a COVID-19 diagnosis at our institution. We matched 2 referent patients to every 1 CD patient. We adjusted for general comorbidities known to be associated with severe COVID-19 including diabetes mellitus, lung disease, obesity, and tobacco use. In addition, we also investigated the presence of celiac-specific comorbidities that may increase the risk for severe COVID-19 including associated type 1 diabetes mellitus, autoimmune hepatitis, rheumatologic conditions and immunosuppressive drugs, common variable immunodeficiency, refractory celiac disease diagnosis, and selective immunoglobulin (Ig) A deficiency.

The primary outcome for our study was COVID-19 infection-related hospitalization. Secondary outcomes included ICU requirement, mortality, and thrombosis. For CD patients we also collected data regarding duodenal biopsy (to confirm diagnosis) and tissue transglutaminase IgA levels. For both CD patients and referents we collected baseline demographics, comorbidities, COVID-19 hospitalization information, and COVID-19 vaccination data. Any patients found to have missing data with regard to CD diagnosis, COVID-19 diagnosis, or demographics were excluded from further analysis.

CD patients were systematically identified by a 2-stage process. Stage 1 involved searching our electronic medical records for patients with documented diagnosis of CD based on International Classification of Diseases codes. International Classification of Diseases, Tenth Revision codes for the initial search included Z78.9 (consumes gluten free), R76.8 (positive autoantibody screening for celiac disease), R85.7 (villous atrophy present on biopsy of small intestine), K90.0 (celiac disease), or K63.89 (chronic diarrhea and villous atrophy). Stage 2 required confirmation of biopsy-proven diagnosis by manual chart review. CD diagnosis was defined as having villous atrophy (eg, Marsh 3) on duodenal biopsy.¹⁶ We also collected data about clinical characteristics of CD at time of COVID-19 infection including time since diagnosis, body mass index, nutritional laboratory results (vitamins and minerals, albumin), and celiac serology (as marker of adherence to the gluten-free diet). Data were collected systematically and stored on a Health Insurance Portability and Accountability Act–compliant database (RedCap) using de-identified entries.

We defined vaccination as either having received 2 doses of the Pfizer or Moderna vaccines or 1 dose of the Johnson & Johnson vaccine. A vaccine booster was defined as any additional dose of vaccine after being fully vaccinated. When assessing for associations between vaccination status and the outcomes, we categorized vaccinated patients on the basis of whether their infection was pre-vaccination or post-vaccination; those who had not yet received the vaccine were categorized as “not vaccinated.” Patients with only 1 dose of the vaccine Pfizer or Moderna were categorized as not vaccinated.

Statistical Analysis

Data were described using median and interquartile range for non-normal continuous variables and frequency (percentage) for categorical variables. Univariate analyses were used to compare cohorts; comparisons for continuous variables were done using t test, and comparisons for categorical variables were done using χ² test. Multivariate analysis with Cox proportional hazard regression analysis was used to determine hazard ratio (HR) for the primary outcome (COVID-19–related hospitalization). Because of the retrospective nature of the study, a sample size estimation was not performed a priori because the study depended on the availability of CD patients with COVID-19 infection diagnosed at our institution during the study period. A post hoc power calculation was performed on the basis of the frequency of the primary outcome (hospitalization) between groups and assumed an alpha of 0.05 and power threshold of 0.8. Ultimately post hoc power analysis demonstrated a power of 84.8%, suggesting that the study was adequately powered.

Statistical analysis was performed using R (version 3.6.2; Vienna, Austria), SAS (version 9.4; Cary, NC), and SPSS (version 29; Armonk, NY) software. A P value <.05 was considered statistically significant.

Results

Between March 1, 2020 and January 1, 2022 there were a total of 110 patients with a biopsy-proven diagnosis of CD and COVID-19 infection. CD patients with COVID-19 were drawn from a population of 171,763 patients diagnosed and treated with COVID-19 at our institution during the study period. Patients with CD were more likely to be female and white, with a mean age of 47 years at COVID-19 diagnosis. After 2:1 matching, 220 referents without CD with COVID-19 infection were included. Distribution of infection incidence and hospitalization can be seen in Figure 1 . Both groups had similar distributions of general comorbidities including obesity (38.2% vs 45.9%), type 2 diabetes mellitus (10.0% vs 14.5%), preexisting lung disease (15.5% vs 11.4%), and tobacco use (24.5% vs 25.5%) (Table 1 ). No patient with CD had coexistent type 1 diabetes mellitus, autoimmune hepatitis, rheumatologic conditions and immunosuppressive drugs use, common variable immunodeficiency, refractory celiac disease diagnosis, and selective IgA deficiency. Patients with CD had a median of 217 months from biopsy diagnosis to time of COVID-19 infection, with a range of 4–726 months; the majority of patients were documented to be following a gluten-free diet (documentation of dietitian interview was available in 73 patients [66.3%]), and tissue transglutaminase IgA was positive in 43 of 93 available (46.2%) at time of COVID-19 infection. Only 1 patient had body mass index below 18, with a body mass index mean of 29.0 (range, 14.2–50.1) at the time of COVID-19 infection. Nutritional laboratory results when available showed no hypoalbuminemia or iron deficiency among CD patients within 3 months of COVID-19 diagnosis.

Figure 1.

COVID-19 incidence curve showing monthly cases and hospitalization in vaccinated and unvaccinated celiac disease and referent patients. CD, celiac disease.

Table 1.

Comparison of Demographics and Comorbidities Between Celiac Disease and Non-Celiac Disease (Referent) Patients With COVID-19 Infection

Factors	Celiac disease patients N = 110	Referent patients N = 220	P value
Age at COVID-19 diagnosis (y)	46.5 [37.0; 60.0]	46.0 [36.0; 60.0]	.047
Sex (%)	Male: 20 (18.2)	Male: 40 (18.2)	1.000
	Female: 90 (82.8)	Female: 180 (82.8)
Race and ethnicity (%)	White: 104 (94.5)	White: 208 (94.5)	1.000
	Black: 3 (2.7)	Black: 6 (2.7)	1.000
	Other/not specified: 3 (2.7)	Other/not specified: 6 (2.7)	1.000
Obesity (%)	42 (38.2)	101 (45.9)	.184
Type 2 diabetes mellitus (%)	11 (10.0)	32 (14.5)	.253
Preexisting lung disease (%)	17 (15.5)	25 (11.4)	.294
Tobacco use (%)	27 (24.5)	56 (25.5)	.844
Vaccinated against COVID-19 (% vaccinated)	71 (64.5)	154 (70.0)	.313
Vaccine brand	Pfizer (%): 50 (70.4%)	Pfizer (%): 99 (64.2%)	.262
	Moderna (%): 18 (25.3%)	Moderna (%): 50 (32.4%)	.185
	Johnson & Johnson (%): 3 (4.2%)	Johnson & Johnson (%): 5 (3.2%)	.657
Boosted against COVID-19 (%)	25 (22.7)	76 (34.5)	.029
Intestinal biopsy positive for CD	110 (100%)	—	NA
Positive IgA-tissue transglutaminase anytime (%)	110 (100)	—	NA

After matching, 27 patients in the CD group (24%) were hospitalized for their COVID-19 infection as opposed to 25 in the referent group (11%) (HR, 2.10; 95% confidence interval [CI], 1.21–3.65; P = .009) (Tables 2 and 3 ). The primary reasons for hospital admission for CD and referents were all COVID-19 related (Table 4 ). Patients in both groups presented most commonly with an acute respiratory condition (63.0% in CD and 76.0% in referents). Presenting respiratory features and diagnoses included dyspnea, hypoxia, pneumonia, and acute respiratory distress syndrome. Supplementary oxygen was required in 17 of 27 (63.0%) hospitalized CD patients (7 vaccinated and 10 unvaccinated) and 21 of 25 (84.0%) hospitalized referents (10 vaccinated and 11 unvaccinated). Supplemental noninvasive oxygen was delivered by nasal cannula in 36 (94.7%) and high flow device in 2 (5.3%). No patient received Paxlovid or oral steroids as outpatient management for COVID-19.

Table 2.

Risk of Hospitalization Between Celiac Disease Patients and Referent Individuals With COVID-19 Infection by Univariate Analysis

Variable	Hazard ratio	95% Confidence interval
Preexisting lung disease	3.20	1.48–6.80
Diabetes mellitus	0.96	0.40–2.28
Obesity	1.41	0.73–2.75
COVID-19 vaccine	0.44	0.24–0.80
Celiac disease	2.54	1.39–4.63
Tobacco use	1.32	0.66–2.65

Table 3.

Risk of Hospitalization Among Celiac Disease Patients and Referent Individuals With COVID-19 Infection by Multivariate Analysis

Variable	Hazard ratio	95% Confidence interval
Preexisting lung disease	2.62	1.40–4.93
Diabetes mellitus	0.883	0.52–2.44
Obesity	0.863	0.49–1.52
COVID-19 vaccine	0.535	0.31–0.93
Celiac disease	2.10	1.21–3.65
Tobacco use	1.29	0.71–2.34

Table 4.

Primary COVID-Related Reasons for Hospitalization and Chief Complaints Among Hospitalized Patients

	Celiac disease patients (N = 27)	Referent patients (N = 25)
Primary reason for hospitalization, n (%)
COVID-19 pneumonia	12 (44.4)	13 (52)
COVID-19 infection	15 (65.6)	12 (48)
Chief complaint at presentation, n (%)
Dyspnea	19 (70.4)	19 (76.0)
Altered mental status	2 (7.40)	0 (0.00)
Chest pain	2 (7.40)	1 (4.00)
Emesis	1 (3.70)	0 (0.00)
Diarrhea	1 (3.70)	0 (0.00)
Syncope	1 (3.70)	1 (4.00)
Fever	1 (3.70)	2 (8.00)
Acute kidney injury	0 (0.00)	1 (4.00)
Abscess	0 (0.00)	1 (4.00)

A total of 71 CD patients were vaccinated (70.4% Pfizer, 25.3% Moderna, and 4.2% Johnson & Johnson), and 154 referent patients were vaccinated (64.2% Pfizer, 32.4% Moderna, and 3.2% Johnson & Johnson). Incomplete vaccination (only 1 dose) with Moderna or Pfizer was observed in 5 CD patients and 7 referent patients. Multivariate analysis showed that vaccination was associated with a lower risk of hospitalization (HR, 0.53; 95% CI, 0.31–0.93; P = .026). Hospitalization did not significantly differ between vaccinated CD patients compared with vaccinated non-CD referent patients (odds ratio, 1.12; 95% CI, 0.35–4.0; P = .79).

Occurrences of secondary outcomes including ICU requirement, mortality, and thrombosis were minimal in both groups (Table 5 ). Four unvaccinated CD patients as opposed to 2 unvaccinated referent patients had ICU requirement. No patients in the CD cohort required mechanical ventilation, whereas 1 hospitalized unvaccinated patient in the referent cohort required it. Last, there were no COVID-19–related mortalities in the CD cohort, whereas there were 2 (1 vaccinated and 1 unvaccinated) in the referent cohort. Neither group had a documented incident thrombosis associated with COVID-19 illness.

Table 5.

Distribution of Outcomes Between Celiac Disease Patients and Non-Celiac Disease (Referent) Patients With COVID-19 Infection

Outcome	Celiac disease patients N = 110	Referent patients N = 220
Hospitalization (%)	27 (24.5)	25 (11.4)
ICU requirement (%)	4 (3.63)	2 (0.91)
Mortality (%)	0 (0)	2 (0.91)

Discussion

The primary findings of this study are that patients with biopsy-proven CD and COVID-19 infection had a higher risk of hospitalization compared with patients without CD after matching for key demographic features and adjusting for general and celiac-specific comorbidity. Vaccination against COVID-19 was associated with a decreased risk of hospitalization in both the CD and referent groups.

This study demonstrates a vaccination effect on mitigation of the risk of hospitalization in CD patients with COVID-19 infection. The main reason for this unique finding and what sets this study apart from prior work in the field is that earlier research did not take into account vaccination status. This study shows significantly different rates of hospitalization among patients with CD depending on vaccination status, with strong evidence for mitigation of hospitalization risk through vaccination. Increased hospitalization risk among CD patients is in contrast to recent studies such as the population-based work conducted by Lebwohl et al¹¹ in Sweden and a large multicenter network study by our group in the United States showing no significant increase in hospitalization risk between CD and non-CD patients with COVID-19 infection.¹² Our present study came to a different conclusion likely owing to incorporating data on vaccination status, which was not evaluated in these earlier studies. Matching for age, gender, and date of COVID-19 diagnosis as well as adjusting for general and celiac-specific comorbidity decreased confounding risk; however, hospitalization risk may be impacted by regional practices for COVID-19 management (our institution is a regional referral center) and selection bias because patients with mild symptoms or self-diagnosis with a home test not seeking medical attention are unlikely to be included in this study. However, regional practices for COVID-19 management and selection bias affect both groups in a similar degree and therefore are unlikely to explain the difference demonstrated in the study. In fact, hospitalization was seen in 18% of CD patients and matched non-CD controls in a U.S. multicenter network study¹²; this is higher than 11% for referents but lower than in our patients with CD of 24% in this study. In a similar multicenter network study, hospitalization risk was as high as 40% in liver transplant patients with COVID-19.⁵ Thus, an additional hypothesis that might explain higher hospitalization risk in our CD cohort is the presence of active disease as suggested by positive tissue transglutaminase in 46% at the time of COVID-19 diagnosis despite following a gluten-free diet and evidence of dietitian instruction in the majority of the patients. With respect to secondary outcomes, we found no increased risk for mortality or severe COVID-19 infection.^{11 , 12} Our congruent findings suggest that CD patients with COVID-19 infection are at least not inherently at greater risk for more severe outcomes. These findings are also supported by a Mendelian randomization study conducted in the United Kingdom biobank cohort.¹⁷

The data regarding CD and COVID-19 are rapidly evolving as new variants emerge. Early in the pandemic, Schiepatti et al conducted research in which they determined baseline characteristics of CD patients infected with COVID-19.¹⁸ Similar to our study, they found that CD patients tended to be younger, white, and female. Furthermore, in their study, all 9 confirmed COVID-19 infected patients did not require hospitalization.¹⁸ Zingone et al conducted a study in Northern Italy investigating outcomes in a CD cohort.¹⁹ In this study, 19 reported flu-like symptoms that were not confirmed to be infected with COVID-19 because of lack of widespread testing at the time. After 1-month follow-up, none of the self-reporting patients went on to require hospitalization for their symptoms. Our research confirms that CD patients with COVID-19 infection did not have greater risk of severe outcomes such as ICU requirement or mortality. In fact, ICU requirement was not observed among vaccinated CD patients.

Pre-pandemic research has demonstrated a greater risk of contracting community-acquired respiratory infections among CD patients and more specifically a greater risk of hospitalization for influenza infection and pneumococcal pneumonia.16, 20 Our findings are consistent with these observations that CD and COVID-19 conveyed an elevated risk of hospitalization related to acute respiratory illness.

We believe that our study has many strengths that add to the current literature. Our electronic medical records’ thorough and clear documentation of vaccination status allowed for clear determination of infection timeline, allowing us to determine whether illness occurred before or after their first vaccination dose. Although the efficacy of vaccination has been demonstrated with most of the available vaccines, the efficacy of vaccination to mitigate hospitalization risk in CD patients demonstrated in this study is limited to vaccines available in the United States including Pfizer, Moderna, and to a lesser degree Johnson & Johnson.²¹ Matching a referent cohort using key demographic features allowed us to minimize possible confounding effects that could arise from such features, and matching also for date of COVID-19 diagnosis allowed us to limit the potential influence of different SARS-CoV-2 strains on outcomes.

The study also had limitations including the retrospective nature of the approach. Another potential limitation is the presence of false-positive COVID-19 results, which are reported to occur in about 0.05% of tests.²² In addition, the higher frequency of false-negative COVID-19 results early in the disease process may reduce inclusion of potential patients with an acute viral illness consistent with COVID-19 infection because the inclusion criteria required a positive test. We also had difficulty determining treatment regimens for hospitalized and especially outpatient individuals with COVID-19 infection, which may have impacted outcomes. However, specific treatments for COVID-19 in the outpatient setting only recently became available in the United States, so this may not be as relevant for our study because of the time period it was conducted²³ (for example, no patient received Paxlovid in either group). Data from electronic health records can be susceptible to documentation errors and omissions. We attempted to mitigate this by ensuring that our CD cohort had a confirmed diagnosis of CD by manually reviewing pathology reports (eg, duodenal biopsy), and the referent cohort lacks a clinical diagnosis of CD after systematic review of the medical record. A potential downside to this approach is that it led to exclusion of some patients with CD without available duodenal biopsy in the medical record. This limitation should be balanced against the benefit of a well-defined biopsy-proven cohort of patients with CD and COVID-19 infection. Data regarding gluten-free diet adherence were inconsistently reported in the charts. Therefore, the size of the risk of COVID-19 infection in CD patients non-adherent to the gluten-free diet could not be fully elucidated. Other research suggests that COVID-19 symptoms and anti–SARS-CoV-2 antibodies were not affected by gluten-free diet adherence.²⁴ Last, vaccination rates are lower in the United States, with 66% of the population being fully vaccinated compared with other regions across the world (eg, 69% in Europe), so the data from this study may not be generalizable to populations with higher vaccination rates.²⁵

Further research that is based on our findings is warranted. Prospective studies that assess COVID-19 outcomes as they relate to gluten-free diet adherence could identify whether CD activity is associated with worse infection outcomes. Parent-reported intentional gluten consumption due to limited gluten-free food availability in the household increased during the pandemic (7.5%).²⁶ Poor compliance to the gluten-free diet increased during the pandemic among 505 CD patients, mostly because of cost and difficult access to gluten-free foods in India for example.25, 27 Because of the rise of “long COVID,”²⁸ it would also be worthwhile to determine whether CD patients are at an increased risk of developing the associated post-infectious sequelae (ie, fatigue, cognitive dysfunction, respiratory difficulty among others).

In conclusion, this study demonstrated a higher risk of hospitalization among biopsy-proven CD patients with COVID-19 infection compared with those without CD. In addition, our study shows a mitigation of the risk of hospitalization among CD patients vaccinated against COVID-19 similar to that seen in vaccinated non-CD patients. Despite this increased risk of hospitalization, there were no significant differences between CD and non-CD patients with respect to ICU requirement, mortality, or thrombosis. Vaccination against COVID-19 should be strongly recommended in patients with CD.