Characteristics of Patients Admitted with and without COVID-19 and Gastrointestinal Bleeding During the Early Pandemic: A Retrospective Review

doi:10.21203/rs.3.rs-3396068/v1

Download PDF

Article

Characteristics of Patients Admitted with and without COVID-19 and Gastrointestinal Bleeding During the Early Pandemic: A Retrospective Review

https://doi.org/10.21203/rs.3.rs-3396068/v1

This work is licensed under a CC BY 4.0 License

Version 1

posted

You are reading this latest preprint version

During the first wave of the pandemic (March 2020-July 2020), it was postulated that there may be an increased risk of gastrointestinal bleeding (GIB) due to COVID-related inflammatory effects on the intestinal mucosa, as well as microthrombi from the hypercoagulable state induced by COVID. We performed a retrospective, multi-center, 1:2 case-control study to review 122 COVID-positive patients and 269 patients without COVID. There were no significant differences between the COVID and non-COVID patients for increased risk of GI bleeding. We concluded that despite the altered pathophysiology of the culprit lesions due to COVID infection, the same standard medical and endoscopic management strategies should be employed.

Health sciences/Gastroenterology

Health sciences/Gastroenterology/Colonoscopy

Health sciences/Gastroenterology/Gastrointestinal diseases

Health sciences/Gastroenterology/Gastrointestinal system

Health sciences/Gastroenterology/Oesophagogastroscopy

When the Coronavirus disease 2019 (COVID) pandemic began in 2020, little was known regarding the systemic and non-respiratory effects of the virus. It was postulated during the first wave of the pandemic (March 2020-July 2020) that there may be an increased risk of gastrointestinal bleeding (GIB) due to COVID-related inflammatory effects on the intestinal mucosa, as well as microthrombi from the hypercoagulable state induced by COVID [1]. Mucosal inflammation and ischemic changes could increase the risk of bleeding throughout the gastrointestinal tract by compromising mucosal integrity. Additionally, the therapeutic interventions for COVID, including the use of corticosteroids and anticoagulants, could increase the risk of bleeding. It is unknown whether acuity of illness due to COVID or other factors, such as medications and co-morbidities, increase the risk of bleeding in patients with COVID as compared to those without COVID. Multiple studies analyzing GI bleeding in COVID patients early in the pandemic illustrated increased rates of gastroduodenal ulcer and rectal ulcer bleeding [2, 3]. These were attributed to critical illness and rectal tube placement, respectively.

Although adequate patient stabilization and endoscopic evaluation within 24 hours of presentation remain the first-line management of GIB, endoscopic evaluation was not routinely performed at the onset of the COVID pandemic. Conservative management strategies, such as supportive transfusions or high-dose proton pump inhibitors (PPI), were often pursued instead due to many factors during the early stages of the pandemic.

The purpose of this study was to further elucidate characteristics of culprit lesions and compare COVID and non-COVID populations with GIB. Prior to the initiation of this study, not many large population-based studies were conducted to analyze potential differences between these two groups. Anecdotal evidence early in the pandemic from our institution indicated duodenal bulb ulcers as the etiology of severe GIB. Little was known regarding the risk factors predisposing COVID-infected patients to GIB versus the non–COVID-infected patients at this point. Additionally, at the time there were little data available that evaluated the severity of GI bleeding in COVID-infected patients. In this study, we retrospectively review charts of patients admitted with GIB with and without COVID infection during the first relative peak and into the start of the second peak of the pandemic (July 2020 to January 2021). The primary aim of this study was to determine the etiology of COVID-related GIB. Secondary aims include determining the risk factors which may contribute to GIB and determining any correlation between COVID and the severity and location of GIB observed in these patients.

Patient Population

This retrospective, multi-center, 1:2 case-control study reviewed COVID-positive patients and patients without COVID who were hospitalized with gastrointestinal bleeding between July 1, 2020 and January 30, 2021. We defined the cases and controls as patients with and without COVID, respectively, with cases defined by a positive RT-PCR for COVID upon chart review. During this period, there were vastly more patients without than with COVID, so we compared the groups with a 2:1 ratio. The non-COVID patients were randomized using the “random” function within Excel, and the first 269 patients were included as the “non-COVID” group in the final analysis. Data were collected from six hospitals within the Houston Methodist network: Houston Methodist Sugarland, Houston Methodist Willowbrook, Houston Methodist Baytown, Houston Methodist Hospital Downtown, Houston Methodist West, Houston Methodist the Woodlands. The patients with COVID-19 had a laboratory confirmed diagnosis using real-time polymerase chain reaction (RT-PCR). This study was approved by the Institutional Review Board at Houston Methodist Hospital, Houston, Texas.

Patients were identified in the medical record software (EPIC) using the diagnosis “GI Bleed” ICD10 codes, refined by “admission” and further refined by timeframe of July 1, 2020-January 30, 2021. We refined our search even further by patients requiring endoscopic evaluation (esophagogastroduodenoscopy [EGD], enteroscopy, video capsule endoscopy, colonoscopy, flexible sigmoidoscopy). Patients who did not have endoscopic evaluation as part of their GI evaluation were excluded from the dataset.

Patient Characteristics

Both the case and control patient charts were evaluated to collect demographic data, clinical history, hospital course, endoscopy reports, and medication use. The characteristics of GI bleeding were distinguished based on the location and etiology of bleeding as recorded in the endoscopy/procedure reports. Therapeutic interventions were also gathered from the procedure reports.

We defined severe GIB as patients admitted with an initial hemoglobin of ≤ 7 g/dL. As many patients in our study had lower (colonic/rectal) or small bowel or a combination of multiple sites of bleeding, we considered the existing scores for predicting GIB severity (Glasgow-Blatchford, Rockall, etc.) to be insufficient, as they pertain to upper GIB only. Critically ill patients were those admitted directly to the intensive care unit (ICU). Anticoagulation (AC) use included use of unfractionated heparin, low molecular weight heparin, warfarin, dabigatran, rivaroxaban, or apixaban. Non-steroidal anti-inflammatory drug (NSAID) use was defined as use of ibuprofen, naproxen, ketorolac, diclofenac, celecoxib, or meloxicam. Antiplatelet agent use was defined as the use of aspirin, thienopyridines (clopidogrel, prasugrel, ticlopidine, ticagrelor), or GPIIb/IIIa inhibitors (abciximab, eptifibatide, and tirofiban).

Statistical Analysis

Patients’ baseline and culprit lesion characteristics were presented as mean ± standard deviation for continuous measures and number (%) for categorical measures by COVID and Non-COVID. Chi-square or Fisher’s exact test for categorical variables and T-test or Mann-Whitney test for continuous variables were used to compare the differences between patients with COVID and those without COVID. Shapiro–Wilk test was used to test normality assumption. The sub-analysis of patients with ulcers noted on endoscopic evaluation within the COVID group was compared between patients with Hgb ≤ 7 g/dL and Hgb > 7 g/dL. Univariable logistic regression analysis was used to examine risk factors for bleeding.

Finally, we analyzed the characteristics of patients requiring intervention for management. We studied these characteristics further via univariable logistic regression analysis to determine any associations between requiring intervention and patient demographics, ulcer lesion type, transfusion requirement, critical illness, medications, and AC. Statistical analyses were performed using Stata V17 (College Station, TX). P-values < 0.05 were considered statistically significant.

Patient Baseline Characteristics

This cohort included 122 COVID patients and 269 non-COVID patients. Overall, the baseline characteristics were similar between the COVID and non-COVID patients (Table 1), but there were a few statistically significant differences. Mean BMI was higher in the COVID patients (29.45 ± 7.52 vs. 27.83 ± 7.59; p = 0.051). A higher proportion of COVID patients were diabetic (50.82% vs. 36.80%; p = 0.009). More non-COVID patients were taking NSAIDs at the time of admission (10.78% vs. 2.46%; p = 0.005). The initial heart rate (in beats per minute) was higher in COVID patients than non-COVID patients at the time of Gastroenterology consultation (86.25 ± 17.72 vs. 81.87 ± 16.37; p = 0.018). COVID patients received more transfusions than non-COVID patients (72.95% vs. 61.71%; p = 0.031).

Table 1

Baseline Patient Characteristics
	Total	Non-COVID	COVID	p-value
	N = 391	N = 269	N = 122
Age (years)	67.26 ± 14.75	67.24 ± 15.88	67.32 ± 11.97	0.96
Sex				0.29
Female	195 (49.87%)	139 (51.67%)	56 (45.90%)
Male	196 (50.13%)	130 (48.33%)	66 (54.10%)
BMI	28.34 ± 7.60	27.83 ± 7.59	29.45 ± 7.52	0.051
HTN	270 (69.05%)	180 (66.91%)	90 (73.77%)	0.17
Diabetes	161 (41.18%)	99 (36.80%)	62 (50.82%)	0.009
Renal Disease^†	102 (26.09%)	64 (23.79%)	38 (31.15%)	0.12
Liver Disease^‡	53 (13.55%)	37 (13.75%)	16 (13.11%)	0.86
NSAIDs	32 (8.18%)	29 (10.78%)	3 (2.46%)	0.005
Initial SBP (mmHg)	129.69 ± 25.94	130.32 ± 24.81	128.35 ± 28.28	0.49
HR (bpm)	83.26 ± 16.91	81.87 ± 16.37	86.25 ± 17.72	0.018
Hgb (g/dL)	8.58 ± 2.77	8.69 ± 2.70	8.33 ± 2.92	0.23
Platelets (x10⁹/L)	226.57 ± 110.77	230.30 ± 113.51	218.42 ± 104.50	0.33
INR	1.56 (1.38)	1.47 (1.09)	1.75 (1.83)	0.100
Transfused	255 (65.22%)	166 (61.71%)	89 (72.95%)	0.031
Data are presented as mean ± standard deviation or n (%). BMI- body mass index; HTN- hypertension; NSAIDs- non-steroidal anti-inflammatory drugs; SBP- systolic blood pressure; mmHg- millimeters of mercury; HR- heart rate; bpm- beats per minute; Hgb- hemoglobin; INR- international normalization ratio. ^†Renal Disease includes a range from chronic kidney disease stage 2 to end-stage renal disease. ^‡Liver Disease comprises a range of chronic liver disease from no fibrosis to decompensated cirrhosis.

Culprit Lesion Characteristics

The characteristics of the culprit lesion—that is, the lesion felt most likely to have caused bleeding (as determined on endoscopy report review by two endoscopists)—were similar between the two groups (Table 2). The culprit lesion was most frequently located in the colon/rectum (29.75%), stomach (29.11%), and small intestine (17.72%) for the non-COVID patients. The COVID patients’ lesions were most frequently in the stomach (26.19%), small intestine (25%), and colon/rectum (22.62%). Both groups had ulcers as the most common etiology of bleeding (non-COVID: 56.55%; COVID: 55%). The next most common lesion in both groups was vascular lesions (excluding gastroesophageal/rectal varices), comprising 18.62% in the non-COVID and 25% in the COVID group. None of the COVID patients had endoscopic evidence of colitis as an explanation for bleeding, whereas five of the non-COVID patients had endoscopic evidence of colitis; however, this was not statistically significant. Most patients in both groups did not require endoscopic intervention (endoclips, band ligation, bipolar cautery, argon plasma coagulation, etc.) for hemostasis. However, there was a nearly significant difference between the groups, with 29% of non-COVID patients and 38.52% of COVID patients requiring intervention (p = 0.061). A higher proportion of COVID than non-COVID patients required mechanical intervention (36.17% vs. 27.27%), but this was not statistically significant. The most frequent intervention type (36.36% of patients) in the non-COVID group was thermal (either bipolar cautery or argon plasma coagulation). Roughly one-third of each group required multiple modalities of intervention for hemostasis. The modalities employed were bipolar cautery, argon plasma coagulation (APC), band ligation, endoclip placement, and hemostatic spray application. Most patients in each group did not have active bleeding at the time of endoscopy. 20.82% non-COVID patients and 29.51% of COVID patients had actively bleeding lesions at the time of endoscopy; this difference was nearly statistically significant (p = 0.061).

Table 2

Culprit Lesion Characteristics
	Total	Non-COVID	COVID	p-value
	N = 391	N = 269	N = 122
Location of Culprit Lesion	n (%)			0.42
Esophagus	27 (11.16%)	15 (9.49%)	12 (14.29%)
Stomach	68 (28.10%)	46 (29.11%)	22 (26.19%)
Small intestine	49 (20.25%)	28 (17.72%)	21 (25.00%)
Colon & rectum	66 (27.27%)	47 (29.75%)	19 (22.62%)
Multiple	32 (13.22%)	22 (13.92%)	10 (11.90%)
Type of Culprit Lesion				0.19
Ulcer	126 (56.00%)	82 (56.55%)	44 (55.00%)
Non-Variceal Vascular^ℓ	47 (20.89%)	27 (18.62%)	20 (25.00%)
Varices	11 (4.89%)	5 (3.45%)	6 (7.50%)
Hemorrhagic Colitis	5 (2.22%)	5 (3.45%)	0 (0.00%)
Multiple	16 (7.11%)	13 (8.97%)	3 (3.75%)
Other^†	20 (8.89%)	13 (8.97%)	7 (8.75%)
Active Bleeding Lesion	92 (23.53%)	56 (20.82%)	36 (29.51%)	0.061
Intervention Required	125 (31.97%)	78 (29.00%)	47 (38.52%)	0.061
Intervention Type				0.68
Mechanical^*	38 (30.65%)	21 (27.27%)	17 (36.17%)
Cautery^¥	42 (33.87%)	28 (36.36%)	14 (29.79%)
Multiple	40 (32.26%)	26 (33.77%)	14 (29.79%)
Other^‡	4 (3.23%)	2 (2.60%)	2 (4.26%)
Data are presented as n (%). ^ℓNon-Variceal Vascular lesions- arteriovenous malformations (AVMs), Dieulafoy lesion, hemorrhoids. ^†Other- friable mucosa, polyp, erosive gastritis, Mallory-Weiss tear, mass lesion, severe esophagitis, mucosal tear. ^*Mechanical-endoclip, variceal ligation. ^¥Cautery- argon plasma coagulation (APC), bipolar cautery. ^‡ Other- Hemospray™ (Cook Medical, USA), epinephrine injection alone.

COVID Patients with Ulcer as Culprit Lesion

Table 3 illustrates a sub-analysis of the COVID patients found to have an ulcer during endoscopy. Of these patients, nearly half were in the ICU (41.86%). However, most patients were not intubated; only 30.23% were on mechanical ventilation. Most patients were on steroids (58.14%), but the majority were not on anticoagulant, antiplatelet or NSAID medications (frequency of use: 34.88%, 39.53%, 2.33%, respectively). Ten of the 43 (23.26%) patients had an actively bleeding ulcer during endoscopy. Seventeen (39.53%) patients required endoscopic intervention for either actively bleeding or high bleeding risk lesions. The most common location of ulcers was in the stomach (32.56%). Six patients had ulcers in multiple locations.

Table 3

COVID Ulcer Patient Characteristics
	Total
	N = 43 (%)
Located in ICU	18 (41.86%)
Intubated	13 (30.23%)
Steroid	25 (58.14%)
Anticoagulant^†	15 (34.88%)
Antiplatelet^*	17 (39.53%)
NSAIDs	1 (2.33%)
Active Bleeding Lesion	10 (23.26%)
Location of Ulcer
Esophagus	5 (11.63%)
Stomach	14 (32.56%)
Duodenal bulb	6 (13.95%)
Small bowel distal to bulb	2 (4.65%)
Colon	6 (13.95%)
Rectum	4 (9.30%)
Multiple	6 (13.95%)
Intervention Required	17 (39.53%)
Intervention Type
Endoclip	8 (47.06%)
Multiple^ℓ	8 (47.06%)
Other^‡	1 (5.88%)
Data are presented as n (%). ICU- intensive care unit; NSAIDs- non-steroidal anti-inflammatory drugs. ^†Anticoagulants include unfractionated heparin, low molecular weight heparin, warfarin, dabigatran, rivaroxaban, and apixaban. ^*Antiplatelets include aspirin, thienopyridines (clopidogrel, prasugrel, ticlopidine, ticagrelor), and GPIIb/IIIa inhibitors (abciximab, eptifibatide, and tirofiban). ^ℓMultiple- any combination of at least two of the following: Endoclip, Hemospray™ (Cook Medical, USA), APC, bipolar probe, epinephrine injection. ^‡Other- Hemospray™ (Cook Medical, USA), epinephrine injection alone.

Severe GIB

Among the COVID patients with GI bleeding, we also analyzed those with severe anemia (designated as Hgb less than 7 g/dL) as a surrogate marker for severity of bleeding (Table 4). Female patients were more likely to have severe anemia than male patients (60% vs. 40%; p = 0.019). The patients without ulcers were more likely to have severe anemia than those with ulcers (62.5% vs. 37.5%; p = 0.018).

Table 4

Comparison of Patient and Culprit Lesion Characteristics in COVID Patients with Severe Anemia
	Total	COVID with Hgb > 7 g/dL	COVID with Hgb ≤ 7 g/dL	p-value
	N = 86	N = 46	N = 40
Age (years)	69.08 ± 10.92	69.39 ± 9.55	68.72 ± 12.44	0.78
Sex				0.019
Female	40 (46.51%)	16 (34.78%)	24 (60.00%)
Male	46 (53.49%)	30 (65.22%)	16 (40.00%)
Anticoagulant	32 (37.21%)	13 (28.26%)	19 (47.50%)	0.066
Intubated	25 (29.07%)	13 (28.26%)	12 (30.00%)	0.86
NSAIDs	3 (3.49%)	2 (4.35%)	1 (2.50%)	1.00
Location of Lesion				0.29
Esophagus	12 (13.95%)	8 (17.39%)	4 (10.00%)
Stomach	22 (25.58%)	11 (23.91%)	11 (27.50%)
Duodenal bulb	11 (12.79%)	4 (8.70%)	7 (17.50%)
Small bowel distal to bulb	10 (11.63%)	3 (6.52%)	7 (17.50%)
Colon	12 (13.95%)	6 (13.04%)	6 (15.00%)
Rectum	8 (9.30%)	6 (13.04%)	2 (5.00%)
Multiple	11 (12.79%)	8 (17.39%)	3 (7.50%)
Ulcer	44 (51.16%)	29 (63.04%)	15 (37.50%)	0.018
Data are presented as mean ± standard deviation or n (%). NSAIDs- non-steroidal anti-inflammatory drugs

Univariable logistic regression analysis was performed on COVID patients with severe GIB (as indicated by Hgb ≤ 7 g/dL) (Table 5). This shows that the patients without ulcers and patients who were female were more likely to have severe GIB. Patients with culprit lesions other than ulcers (OR 2.84, 95% CI: 1.18–6.83, p = 0.019) and female COVID patients (OR 2.81, 95% CI: 1.17–6.76, p = 0.021) were significantly associated with severe gastrointestinal bleeding. Table 6 shows the majority of patients with ulcers did not require endoscopic intervention (70.27% vs. 36.73%; p = 0.002). There were no statistically significant differences in all other patient and lesion characteristics between intervention and no intervention. Univariable analysis shows that COVID patients with culprit lesions other than ulcers had four times higher odds of not requiring endoscopic intervention than those with ulcers (OR 4.07, 95% CI: 1.63–10.15, p = 0.003) (Table 7).

Table 5

Association with Severe Gastrointestinal Bleeding from Univariable Logistic Regression Analysis
	Hgb ≤ 7 g/dL
Variables	No (n = 46)	Yes (n = 40)	OR	95% CI	p-value
Lesion other than Ulcer	17 (36.96%)	25 (62.50%)	2.84	(1.18, 6.83)	0.019
Age (years)	69.39 ± 9.55	68.72 ± 12.44	0.99	(0.96, 1.03)	0.78
Female Sex	16 (34.78%)	24 (60.00%)	2.81	(1.17, 6.76)	0.021
Anticoagulants	13 (28.26%)	19 (47.50%)	2.30	(0.94, 5.61))	0.068
Intubated	13 (28.26%)	12 (30.00%)	1.09	(0.43, 2.76)	0.859
NSAIDs	2 (4.35%)	1 (2.50%)	0.56	(0.05, 6.47)	0.645
Data are presented as mean ± standard deviation or n (%). NSAIDs- non-steroidal anti-inflammatory drugs

Table 6

Comparison of Patient and Lesion Characteristics in COVID Patients Requiring Endoscopic Intervention
	Total	No Intervention	Intervention	p-value
	N = 86	N = 37	N = 49
Age (years)	69.08 ± 10.92	66.68 ± 11.07	70.90 ± 10.57	0.076
Sex				0.33
Female	40 (46.51%)	15 (40.54%)	25 (51.02%)
Male	46 (53.49%)	22 (59.46%)	24 (48.98%)
Anticoagulant	32 (37.21%)	15 (40.54%)	17 (34.69%)	0.58
Intubated	25 (29.07%)	9 (24.32%)	16 (32.65%)	0.40
NSAIDs	3 (3.49%)	1 (2.70%)	2 (4.08%)	1.00
Location of Ulcer				0.11
Colon	12 (13.95%)	4 (10.81%)	8 (16.33%)
Duodenal bulb	11 (12.79%)	3 (8.11%)	8 (16.33%)
Esophagus	12 (13.95%)	6 (16.22%)	6 (12.24%)
Multiple	11 (12.79%)	3 (8.11%)	8 (16.33%)
Rectum	8 (9.30%)	6 (16.22%)	2 (4.08%)
Small bowel distal to bulb	10 (11.63%)	2 (5.41%)	8 (16.33%)
Stomach	22 (25.58%)	13 (35.14%)	9 (18.37%)
Ulcer	44 (51.16%)	26 (70.27%)	18 (36.73%)	0.002
Transfuse	66 (76.74%)	27 (72.97%)	39 (79.59%)	0.47
Data are presented as mean ± standard deviation and n (%). NSAIDs- non-steroidal anti-inflammatory drugs

Table 7

Association with COVID Patients Requiring Endoscopic Intervention from Univariable Logistic Regression Analysis
Variables	No Intervention (n = 37)	Intervention (n = 49)	OR	95% CI	p-value
Univariate models
Age (years)	66.68 ± 11.07	70.90 ± 10.57	1.04	(1.00, 1.08)	0.084
Non-Ulcer Lesion (vs. ulcer)	26 (70.27%)	18 (36.73%)	4.07	(1.63, 10.15)	0.003
Male (vs. female)	22 (59.46%)	24 (48.98%)	0.66	(0.28, 1.55)	0.336
Transfused (vs. not)	27 (72.97%)	39 (79.59%)	1.44	(0.53, 3.94)	0.473
Anticoagulants (vs. none)	15 (40.54%)	17 (34.69%)	0.78	(0.32, 1.88)	0.579
Intubated (vs. not)	9 (24.32%)	16 (32.65%)	1.51	(0.58, 3.94)	0.401
NSAIDs (vs. none)	1 (2.70%)	2 (4.08%)	1.53	(0.13, 17.57)	0.732
Data are presented as mean ± standard deviation and n (%). NSAIDs- non-steroidal anti-inflammatory drugs

This study analyzes the clinical and culprit lesion characteristics of patients admitted with COVID and GIB during the beginning of the global pandemic. Most gastroenterology studies produced during the initial phases of the COVID pandemic focused on gastrointestinal symptoms and mortality data regarding GIB [2–7]. However, few studies have compared COVID patients with GIB to the non–COVID-infected population admitted with GIB, and few have evaluated the severity of GIB outside the upper gastrointestinal tract. Overall, our patient population was similar to previously published populations- both the COVID and non-COVID groups with a few exceptions. The most frequent location of the culprit lesion was in the stomach for COVID patients whereas the colon and rectum were the most frequent, closely followed by stomach for the non-COVID population. Ulcers were the most common lesion type for both groups but did not require as many endoscopic interventions as the non-ulcer lesions. Females and presence of culprit lesions other than ulcers at endoscopy had a significant association with severe GIB. Overall, findings from the COVID and non-COVID groups in our study were similar.

Our patient population was similar to previously published studies on GIB in COVID patients [2–4]. The mean age was within the 7th decade, and most patients had hypertension, diabetes, and were overweight. Although few other studies have compared COVID and non-COVID GIB populations, Gonzalez, et al. had a 249-patient control group comprising non-COVID GIB patients [5]. Their baseline characteristics closely resemble those of our non-COVID patients, except that our cohort had a higher proportion of female patients and a higher frequency of diabetes, renal disease (comprising a range of disease along the spectrum from chronic kidney disease 2 to end stage renal disease), as well as variation in the presentation vital signs and baseline lab values. Our COVID patients also had a significantly higher BMI than the non-COVID GIB population, whereas the population in the Gonzalez study did not have a significant difference in rates of obesity (however BMI was not quantified in this study) [5]. Although there was wide variation in the rate of diabetic patients and in significant differences between case and controls in previous studies, we had higher rates of diabetic patients in both the COVID (50.82%) and non-COVID (36.80%) groups than most studies [3, 5–7]. This may be due in part to geographical location, socioeconomic and cultural differences between the cohorts included in other studies as compared to ours. In our study, NSAID use was significantly higher among the non-COVID than the COVID patients: 10.78% vs 2.46%(p = 0.005). The lower NSAID use in COVID patients in general may be due to advice based on anecdotal evidence early in the pandemic to avoid NSAIDs (specifically ibuprofen), as they were initially thought to increase entrance of the COVID virion into cells by upregulating the angiotensin-converting enzyme 2 (ACE2)-receptors. This hypothesis was later disproven by multiple studies [8–11]. Heart rate was significantly higher in COVID patients: 86 bpm vs. 81 bpm in non-COVID patients (p = 0.018). This difference between groups may be explained by the systemic inflammatory effect spurred by COVID infection. We also found that more COVID patients had blood transfusions than non-COVID patients. This may be due to mechanical consumption of coagulation factors/platelets, blood cells or disseminated intravascular coagulation (DIC) as a manifestation of critical illness. Overall, ulcers were the most common etiology for bleeding in both groups. This finding is similar to prior studies [2, 3, 5–7]. The most common culprit lesion location in COVID patients was the stomach. The most common location of the culprit lesion in the non-COVID patients in our study was colon and rectum, closely followed by the stomach. Among the COVID patients with ulcers, most were not located in the ICU, were not intubated and were not on NSAIDs, AC or antiplatelet medication. However, most patients were on steroids (58.14%). Steroids were a mainstay of COVID therapy early in the pandemic, particularly in patients with COVID pneumonia. Steroid utilization may explain the etiology of the ulcers, as steroids are a well-known risk factor for ulcer formation [12–15]. Additionally, the underlying inflammatory conditions of COVID itself, or perhaps disordered coagulation from COVID or critical illness, may have potentiating effects on preexisting lesions. COVID is known to enter cells through the ACE2-receptor, which is found in high densities within lung parenchyma, as well as in enterocytes [12]. Autopsy and bowel resection specimens from patients early in the pandemic who died from complications of COVID showed evidence of chronic inflammation (interstitial edema within the lamina propria, infiltrating plasma cells and lymphocytes) and ischemic changes (endothelialitis, coagulative necrosis, microthrombi) throughout the intestine [1, 16–18].

More COVID patients had active bleeding during endoscopy than non-COVID patients in this study (29.51% vs. 20.82%, respectively; p = 0.061). The most likely explanation for the lack of significant difference is that endoscopy was delayed or deferred in COVID patients unless they had outward signs of active bleeding (bloody nasogastric tube output, melena, hematochezia, etc.). This was due to fears of contagion, concern for personal protective equipment overutilization, and the logistics of arranging inpatient endoscopy with limited personnel during the early stages of the pandemic.

Our analysis of severely anemic COVID patients (designated as having a hemoglobin ≤ 7 g/dL) as a surrogate marker for severe bleeding found that many of these patients were female and had culprit lesions other than ulcers. However, we did not compare deviation of admission Hgb from baseline Hgb due to lack of available data.

The strengths of our study include the large size of cohort, length of study, wealth of endoscopic data, and multicenter source of data. Most other studies had < 100 COVID GIB patients, or they compared COVID patients with and without GIB. Most other studies also only examined a short period of time: 1–3 months in most cases. However, it is worth noting that most of these studies were published when little was known about the effects of COVID, so publishing data was urgent. Our retrospective review offers the advantage of a historic perspective. The longer time-frame of the study and multiple centers included afforded a larger cohort. We also had the advantage of uniform endoscopic reports available through the same electronic medical record.

The limitations of our study include variability in endoscopist management and therapeutic strategies, as the patients in our cohort were seen by many different endoscopists across multiple centers. The COVID GIB population may have had a higher rate of bleeding lesions than observed in this study—many critically ill COVID patients were simply too sick to undergo endoscopic evaluation. Although we did not specifically analyze endoscopy timing, this may have varied considerably between the groups. As a retrospective review, this study was limited in observing differences in management and certain outcomes (mortality). Mortality was most likely higher in the COVID group due to other variables, namely respiratory and multiple organ failure. The timespan of the study, between the first two “waves” of the pandemic, may have affected our findings. Rapidly changing recommendations regarding COVID management may have impacted management strategies and endoscopy logistics, and the differing strains of the virus may have had different clinical impacts. Finally, the implementation of the COVID-19 vaccine, beginning in December 2020, and its impacts on the severity of illness may have affected outcomes in our study population.

This study serves as a reminder that even in unfamiliar territory, such as a global pandemic, the same basic principles of management apply. As this study reinforces, there were no significant differences between the COVID and non-COVID patients. Even if the pathophysiology of the culprit lesions was altered by COVID infection, the basic medical and endoscopic management strategies should be employed.

Disclosures:

The authors have no conflicts of interest or funding to disclose.

Data Availability

All data generated or analyzed during this study are available from the corresponding author on reasonable request. The published contains the results of the data analysis, respresented in the form of tables.

Aslan AT, Şimşek H. Clinical Features and Pathophysiological Mechanisms of COVID-19-associated Gastrointestinal Manifestations. Euroasian J Hepatogastroenterol. 2021;11(2):81–86.
Martin TA, Wan DW, Hajifathalian K, et al. Gastrointestinal bleeding in patients with coronavirus disease 2019: a matched case-control study. American Journal of Gastroenterology, vol. 115, no. 10, pp. 1609–1616, 2020.
Trindade AJ, Izard S, Coppa K, et al. Gastrointestinal bleeding in hospitalized COVID-19 patients: a propensity score matched cohort study. Journal of Internal Medicine, vol. 289, 2020.
Makker J, Mantri N, Patel HK, et al. The Incidence and Mortality Impact of Gastrointestinal Bleeding in Hospitalized COVID-19 Patients. Clin Exp Gastroenterol. 2021;14:405–411. Published 2021 Oct 8.
González González R, Jacob J, Miró Ò, et al. Incidence, clinical characteristics, risk factors, and outcomes of upper gastrointestinal bleeding in patients with COVID-19. Journal of Clinical Gastroenterology. 2020;25.
Mauro A, De Grazia F, Lenti MV, et al. Upper gastrointestinal bleeding in COVID-19 inpatients: Incidence and management in a multicenter experience from Northern Italy. Clin Res Hepatol Gastroenterol. 2021;45(3):101521.
Abulawi A, Al-Tarbsheh A, Leamon A, et al. (July 03, 2022) Clinical Characteristics of Hospitalized COVID-19 Patients Who Have Gastrointestinal Bleeds Requiring Intervention: A Case-Control Study. Cureus 14(7): e26538.
Kragholm K, Torp-Petersen C, Fosbol E. Non-steroidal anti-inflammatory drug use in COVID-19. The Lancet Rheumatology, Volume 3, Issue 7, e465 - e466.
Kushner P, McCarberg BH, Grange L, et al. The use of non-steroidal anti-inflammatory drugs (NSAIDs) in COVID-19. npj Prim. Care Respir. Med. 32, 35 (2022).
Moore N, Bosco-Levy P, Thurin N et al. NSAIDs and COVID-19: A Systematic Review and Meta-analysis. Drug Saf 44, 929–938 (2021).
Reese JT, Coleman B, Chan L et al. NSAID use and clinical outcomes in COVID-19 patients: a 38-center retrospective cohort study. Virol J 19, 84 (2022).
Narum S, Westergren T, Klemp M. Corticosteroids and risk of gastrointestinal bleeding: a systematic review and meta-analysis. BMJ Open 2014;4:e004587.
Carpani de Kaski M, Rentsch R, Levi S, et al. Corticosteroids reduce regenerative repair of epithelium in experimental gastric ulcers. Gut 1995; 37: 613–616.
Luo JC, Shin VY, Liu ES, et al. Non-ulcerogenic dose of dexamethasone delays gastric ulcer healing in rats. J Pharmacol Exp Ther 2003; 307: 692–698.
Luo JC, Shin VY, Liu ES, et al. Dexamethasone delays ulcer healing by inhibition of angiogenesis in rat stomachs. Eur J Pharmacol 2004; 485: 275–281.
Caramaschi S, Kapp ME, Miller SE, et al. Histopathological findings and clinicopathologic correlation in COVID-19: A systematic review. Modern Pathology, (2021). 34(9), 1614–1633.
Gupta A, Madhavan MV, Sehgal K, et al. Extrapulmonary manifestations of COVID-19. Nature Medicine. (2020). 26(7), 1017–1032.
Vanella G, Capurso G, Burti C, et al. Gastrointestinal mucosal damage in patients with COVID-19 undergoing endoscopy: an international multicentre study. BMJ Open Gastroenterology 2021;8:e000578.

No competing interests reported.

Download PDF

Version 1

posted

You are reading this latest preprint version

Characteristics of Patients Admitted with and without COVID-19 and Gastrointestinal Bleeding During the Early Pandemic: A Retrospective Review

Status:

Version 1

Abstract

Introduction

Methods

Patient Population

Patient Characteristics

Statistical Analysis

Results

Patient Baseline Characteristics

Culprit Lesion Characteristics

COVID Patients with Ulcer as Culprit Lesion

Severe GIB

Discussion

Declarations

References

Additional Declarations

Status:

Version 1