Risk factors for gastric mucosa lesion in critically ill patients undergoing endoscopy for percutaneous gastrostomy: a case-control study

Abstract

Background

Pharmacologic prophylaxis for gastric ulcer is commonly prescribed in patients hospitalized in the intensive care unit (ICU). The aim of the study was to assess the current prevalence and risk factors for gastric mucosa lesion in ICU patients receiving standard pharmacologic prophylaxis undergoing endoscopy for percutaneous gastrostomy implantation.

Methods

Patients hospitalized in the mixed medical-surgical ICU undergoing percutaneous endoscopic gastrostomy (PEG) were analyzed. We excluded patients receiving either no or high doses of intravenous proton pump inhibitor (PPI), only patients receiving standard doses of PPI were included. Data retrieved from the electronic medical records included: demographics, risk factors for gastric mucosa lesion (use of stimulants, comorbidities, medications, treatment methods in the ICU, laboratory derangements) and endoscopic findings. The study compared a group of patients with gastric mucosa lesions (cases) vs. patients without gastric mucosa lesions (controls). Inter-group comparisons between cases and controls were performed. Depending on the type of distribution continuous variables were assessed using two-sample t-test or Mann-Whitney test, whereas categorical variables with Chi-squared or Fisher exact test. Odds ratio (OR) with 95% confidence interval (95% CI) were calculated. Statistical significance was assumed at p < 0.05.

Results

Patients with no prophylaxis (n = 8) or receiving high doses (> 40 mg per day) of proton pump inhibitor (n = 2) were excluded. There were 182 patients receiving standard intravenous dose of PPI, 63 (34.6%) women and 119 (65.4%) men, with median age 61.5 (interquartile range IQR 46.0–70.0) years. Majority of patients (n = 169, 92.9%) were receiving pharmacological prophylaxis for venous thromboembolism. There were 53 (29.1%) patients with gastric mucosa lesion. The only risk factor that was significantly different between cases and controls was history of gastric ulcer (p = 0.04) with OR 3.8 (95% CI 1.1–12.5; p = 0.03).

Conclusions

Majority of various risk factors for gastric ulceration may not predict gastric mucosa lesion in ICU patients receiving standard pharmacological prophylaxis undergoing endoscopy for PEG implantation. We found that history of gastric ulcer may be a risk factor for gastric ulceration in the ICU patients. Patients with history of gastric ulcer might benefit from higher than standard doses of anti-ulcer medication when hospitalized in the ICU.

Background

Epidemiological data indicate that gastric ulceration is present in 5–10% of the population, with the annual incidence of 0.1–0.19% [1, 2]. Lack of prevention of gastric ulcer may lead to gastrointestinal bleeding. The standard treatment to prevent stress ulcers in the intensive care unit (ICU) is the use of proton pump inhibitor (PPI) or histamine H₂ receptor antagonist (H₂RA) [3–5]. Proton pump inhibitors are first choice anti-ulcer medications for patients treated with nonsteroidal anti-inflammatory drugs. They are also used to prevent bleeding from peptic ulcers. The standard dose of pantoprazole is 40 mg once a day. Pharmacological gastric ulcer prophylaxis with PPI is commonly prescribed in patients hospitalized in ICUs [6, 7], however it is associated with increased risk of ventilator-associated pneumonia [8] and C. difficile colitis [9]. Pharmacological stress ulcer prophylaxis should be used in patients in whom major risk factors of stress ulcer are present: mechanical ventilation, coagulopathy, traumatic brain injury, major burn [10]. Patients without these risk factors, who receive enteral nutrition, may benefit from discontinuing pharmacological stress ulcer prophylaxis.

Percutaneous endoscopic gastrostomy (PEG) is a procedure performed in patients requiring long-term enteral nutrition, who cannot be fed orally due to swallowing disorders, head and neck tumors, or other chronic conditions. It is an artificial access route to the stomach created for internal nutrition. The patient is anesthetized locally or generally for this procedure and a gastrostomy tube is inserted through the skin of the abdominal wall into the lumen of the stomach. It is a safe and effective way of providing required alimentation. PEG significantly improves patient’s quality of life by eliminating the need for nasogastric (NG) tube. It is safe in patients with dysphagia, following stroke or other cerebrovascular diseases. The risk of aspiration pneumonia and gastroesophageal reflux is lower when patients are fed via PEG compared to NG tube [11, 12].

There are many studies reporting gastrointestinal bleeding in the ICU patients [13–16]. Studies reporting gastroscopic images in the ICU patients are limited [17]. Standard evaluation of the upper gastrointestinal mucosa during gastroscopy preceding PEG implantation allows for an objective assessment of abnormalities and their severity.

The aim of the study was to assess the current prevalence and risk factors for gastric mucosa lesion in the ICU patients receiving standard pharmacological stress ulcer prophylaxis undergoing endoscopy for PEG implantation.

Methods

This was a single-center case-control study carried out in a mixed medical-surgical ICU in a large academic medical center. The analyzed period was January 2019 to December 2023. The study population comprised consecutive patients hospitalized in the local ICU who underwent percutaneous endoscopic gastrostomy (PEG) implantation during their hospitalization in the ICU. Patients who were not receiving pharmacological prophylaxis (no-PPI) or who were treated with high dose PPI (pantoprazole > 40 mg per day, high-PPI) were excluded, only patients receiving standard doses of PPI (pantoprazole 40 mg IV per day) were included. Clinical and endoscopic data were retrieved from electronic medical records (AMMS, Asseco Medical Management Solutions, Poland).

The retrieved data included the following: general demographic and clinical data, comorbidities, clinical aspects of hospitalization in the ICU, laboratory parameters, gastroscopic findings. General demographic and clinical data included: age, sex, body mass index (BMI), length of ICU stay from ICU admission to PEG implantation. Severity of disease was estimated using appropriate classification systems: Acute Physiology and Chronic Health Evaluation II (APACHE II), Simplified Acute Physiology Score (SAPS II), and Sequential Organ Failure Assessment (SOFA). Among comorbidities special interest was given to diseases that could potentially increase the risk of bleeding: chronic kidney disease, acute kidney injury, chronic liver disease, acute liver injury. Stimulants that were of interest in the context of gastric mucosa lesion were the following: nicotine use (light smoker < 20 cigarettes/day; heavy smoker ≥ 20 cigarettes/day; nicotine exposure in pack-years) and alcohol use (allowable consumption; alcoholic). Clinical aspects of hospitalization in the ICU were as follows: use of mechanical ventilation, presence of sepsis, norepinephrine (NE) infusion (dose in µg/kg/min.), use of antiplatelet medications (type and dose), use of anticoagulants (type and dose), use of nonsteroidal anti-inflammatory drugs (NSAIDs) (type and dose) and corticosteroids (type and dose equivalent for hydrocortisone). Laboratory parameters included complete blood count (hemoglobin, platelets) and standard laboratory tests of coagulation (international normalized ratio, activated partial thromboplastin time aPTT, fibrinogen concentration). Gastroscopic examinations were performed by 2 experienced gastroenterologists and included information regarding abnormalities in the gastric mucosa such as: petechial hemorrhages, erosion, ulceration, bleeding. Location of lesions was also recorded.

All statistical analyses were performed using a licensed statistical software (18.0 Basic Edition, Stata, StataCorp LLC, College Station, USA). Continuous variables were presented as means ± standard deviation (SD) or medians (Me) and interquartile range (IQR) depending on the type of distribution. The type of the distribution was investigated by examining a histogram, and additionally using Skewness/Kurtosis or Shapiro-Wilk test. Categorical variables were presented as frequency and percentage. Data were analyzed using appropriate statistical methods to determine the risk factors for gastric mucosa damage. Differences between cases and controls for continuous variables were investigated with two-sample t-test or Mann-Whitney test, whereas differences for binary variables were investigated with Chi-square or Fisher exact test. Potential risk factors for gastric mucosa damage were analyzed using logistic regression with odds ratio (OR) and 95% confidence interval (95% CI). Multivariable analysis of potential risk factors for gastric mucosa lesion was performed using logistic regression. Predictive value of potential risk factors for gastric ulceration was examined by drawing Receiver Operating Characteristic curve and calculating area under the curve (AUROC). Power of the study was calculated. Significance was set at p < 0.05.

The statistical power of the study, calculated for one of the most significant risk factors for gastric mucosal lesions (NSAIDs), was 0.12.

The study was conducted according to the principles expressed in the Declaration of Helsinki. Due to retrospective character of the study, the Bioethics Committee of Medical University of Silesia decided that the study does not require an ethical review and waived the requirement for an informed consent (Ethics Committee of Medical University of Silesia in Katowice decision number: PCN/CBN/0022/KB/292/21).

Results

Following exclusion of patients not receiving pharmacological ulcer prophylaxis (n = 8) or receiving high doses of PPI (n = 2), the final study group comprised 182 patients. The median age in the study group was 61.5 (IQR 46.0–70.0) years. There was not significant difference in age between men and women (p = 0.24). There was 63 (34.6%) women and 119 (65.4%) men in the study group. The median time to PEG implantation since ICU admission was 24 (IQR 17–35) days. The indication for PEG insertion was need for long-term enteral alimentation in patients unable to be fed orally due to neurological impairment or presence of a tracheostomy tube. Out of 182 patients 37 (20.3%) died in the ICU. Majority of patients (86.8%) were receiving enteric nutrition before PEG implantation. There were 53 (29.1%) patients who presented with gastric mucosa lesions (Fig. 1).

Fig. 1.

Endoscopic features of gastric mucosa lesion

The location of gastric mucosa lesions is presented in Table 1.

Table 1.

Location of gastric mucosa lesions

	Location of gastric mucosa lesion (number of cases)
Type of lesion	Cardia	Fundus	Body	Greater curvature	Lesser curvature	Pylorus
Petechial hemorrhages	6	7	7	4	4	1
Erosion	5	4	8	3	1	0
Ulceration	3	2	3	2	0	1
Bleeding	3	0	2	1	1	0

The particular types of lesions were present in more than one location in the following numbers of patients: petechial hemorrhages – 8 out of 25 patients, erosions – 8 out of 19 patients, ulceration – 0 out of 17 patients, and bleeding – 2 out of 8 patients.

Clinical characteristics of the cases and controls are presented in Table 2.

Table 2.

Clinical characteristics in the patient population

Clinical variables	Cases (n = 53)	Controls (n = 129)	P-value	OR (95% CI
Age, Me (IQR) [years]	63 (55–70)	61 (45–70)	0.56	1.00 (0.98–1.03)
Male sex [%]	67.9	64.3	0.64	1.17 (0.59–2.32)
Time to PEG, Me (IQR) [days]	24 (17–33)	23 (17–36)	0.94	0.99 (0.98–1.02)
ICU mortality [%]	20.7	20.2	0.93	1.04 (0.47–2.29)
Body mass index, Me (IQR) [kg m− 2]	26.2 (24.5–33.1)	26.1 (22.9–29.0)	0.15	1.07 (0.99–1.15)
Gastric ulcer history [%]	13.2	3.9	0.04	3.77 (1.14–12.49)
Chronic kidney disease [%]	7.6	3.9	0.29	2.02 (0.52–7.85)
Acute kidney injury [%]	24.5	21.7	0.70	1.17 (0.55–2.49)
Chronic liver disease [%]	3.8	2.3	0.63	1.65 (0.27–10.15)
Acute liver injury [%]	3.8	0.8	0.20	5.02 (0.44–56.58)
Diabetes [%]	22.6	18.6	0.54	1.28 (0.59–2.80)
Coronary artery disease [%]	9.4	13.2	0.62	0.69 (0.24–1.97)
Nicotine exposure [%]: -non smoker -light smoker (< 20 cigarettes/day) -heavy smoker (≥ 20 cigarettes/day)	83.0 9.4 7.6	88.4 5.4 6.2	0.51	0.76 (0.15–3.90)
Alcohol exposure [%]: -abstainer -moderate consumption -alcoholic	83.0 7.6 9.4	83.7 3.1 13.2	0.34	0.93 (0.57–1.52)
Sepsis [%]	17.0	17.8	1.00	0.94 (0.40–2.20)
APACHE II, Me (IQR) [points]	17.0 (14.0–21.5)	17.0 (14.0–21.0)	0.66	1.02 (0.96–1.07)
SAPS II, Me (IQR) [points]	46.0 (39.0–54.0)	44.0 (34.5–53.0)	0.29	1.01 (0.99–1.04)
SOFA, Me (IQR) [points]	7.0 (5.0–9.0)	6.0 (4.0–8.0)	0.19	1.11 (0.99–1.24)

APACHE II – Acute Physiology and Chronic Health Evaluation II, CI – confidence interval ICU – intensive care unit, IQR – interquartile range, Me – median value, OR – odds ratio, PEG – percutaneous endoscopic gastrostomy, SAPS II – Simplified Acute Physiology Score II, SOFA – Sequential Organ Failure Assessment. In bold statistically significant differences

There were no statistically significant differences between cases and controls, with the exception of gastric ulcer history – the percentage among cases was higher than among controls (13.2 vs. 3.9%, p = 0.04). Logistic regression was performed to analyze history of gastric ulcer as a predictor of mucosa damage. Odds ratio for history of peptic ulcer was 3.77 (95% CI 1.14–12.49) (p = 0.03). Value of history of gastric ulcer in prediction of gastric mucosa damage using AUROC was poor (AUROC = 0.55). When only ulcers and bleeding were taken into consideration as the outcome, the difference in prevalence of history ulcer disease between patients with and without this severe mucosa lesion was 58.3% vs. 8.2% (p < 0.001), respectively.

Treatment modalities among cases and controls are presented in Table 3.

Table 3.

Treatment modalities in the patient population

Treatment modality	Cases (n = 53)	Controls (n = 129)	P-value	OR (95% CI)
Mechanical ventilation [%]	84.9	78.3	0.41	1.56 (0.66–3.69)
CRRT [%]	1.9	7.7	0.18	0.23 (0.03–1.83)
Norepinephrine dose, Me (IQR) [µg/kg/min]	0.05 (0.00–0.10)	0.00 (0.00–0.07)	0.05	7.51 (0.55–102.93)
Aspirin use [%]	30.2	29.5	1.00	1.04 (0.51–2.08)
Low-molecular-weight heparin use [%]: -enoxaparin -dalteparin	41.5 52.8	43.4 48.8	0.87	1.17 (0.70–1.97)
NSAID use [%]	7.6	3.9	0.29	2.02 (0.52–7.85)
Corticosteroid use [n (%)]	7.6	19.4	0.07	0.34 (0.11–1.03)
Corticosteroid dose, Me (IQR) [mg hydrocortisone equivalent]	78.3 (50.0–160.0)	50.0 (50.0–106.7)	0.80	1.00 (0.99–1.01)

CI – confidence interval, CRRT – continuous renal replacement therapy, IQR – interquartile range, Me – median value, NSAID – non-steroidal anti-inflammatory drug, OR – odds ratio

There were no statistically significant differences between cases and controls among treatment modalities. Close to statistical significance was difference between NE dose, with higher doses among cases (p = 0.05).

Laboratory parameters in cases and controls are presented in Table 4.

Table 4.

Laboratory parameters in the patient population

Parameter, Me (IQR)	Cases (n = 53)	Controls (n = 129)	P-value	OR (95% CI)
Hemoglobin [g L− 1]	99.0 (78.0–114.0)	95 (83.0–108.5)	0.43	1.06 (0.92–1.22)
Platelets [x 103 µL− 1]	322.0 (241.0–414.0)	323.5 (229.5–404.0)	0.89	0.99 (0.99–1.00)
INR	1.17 (1.11–1.29)	1.16 (1.09–1.26)	0.48	4.96 (0.68–36.27)
aPTT [s]	33.7 (29.5–38.9)	35.5 (28.3–41.7)	0.36	0.98 (0.96–1.01)
Fibrinogen [mg dL− 1]	506.5 (397.0–658.5)	639.5 (414.0–792.0)	0.11	0.99 (0.99–1.00)

aPTT – activated partial thromboplastin time, CI – confidence interval, INR – International Normalized Ratio, IQR – interquartile range, Me – median value, OR – odds ratio

There were no statistically significant differences between cases and controls among laboratory parameters.

We performed multivariable analysis of potential risk factors for gastric mucosa lesion in the population of critically ill patients (mechanical ventilation, coagulopathy, sepsis, catecholamines, renal failure, NSAIDs), however the fitness of the multivariable model was always worse than single variable (adjusted R² lower than R²).

Discussion

Intensive care unit patients receiving standard ulcer prophylactic pharmacotherapy, undergoing percutaneous endoscopic gastrostomy, are a specific group of patients.

Literature provides a vast perspective on PEG implantation related complications [18], its relevance in the ICU patients [19], as well as the need for pharmacologic prophylaxis [20, 21].

In our study gastric mucosa lesions were defined based on endoscopic finding. Four types of lesions were taken into consideration: erosions, petechial hemorrhages, ulcers, and bleeding. Patients with one or more of these lesions were categorized as a case. Gastroscopy included endoscopic evaluation of esophagus, stomach and duodenum. Almost one third of analyzed patients developed damage to the gastric mucosa despite pharmacological prophylaxis. Such outcome is lower than in current literature that quotes the prevalence of acute gastrointestinal injuries in critically ill patients at 40% [22]. This relatively high prevalence highlights the importance for re-evaluation of current preventive protocols in the ICU patients.

Based on the results of our study, a risk factor for gastric ulceration worth taking into consideration is history of ulcer disease. Patients with a history of ulcer disease may be at increased risk of developing gastric mucosa lesion. It may be due to several factors. Ulcer disease often results in chronically compromised mucosal barrier making it more vulnerable to acidic damage [23]. Previous ulcers prevents the tissue from healing due to inflammation caused by cytokines (e.g. tumor necrosis factor α, interleukin 1β) [24]. Healing of ulcers often results in fibrosis and scarring what makes tissue more prone to re-injury compared to healthy tissue [25]. Recently use of PPIs increases yearly [26]. While medication has its adverse effects, its use is necessary in patients with high risk of gastric ulceration [27]. Positive history of ulcer disease identifies a group of patients requiring special attention while prescribing PPIs. The results of our study also stress the need for thorough history taking. Lack of association between standard risk factors and gastric ulceration in the evaluated group of patients suggests that the current regimen of pharmacology is sufficient for the majority of ICU patients undergoing PEG implantation. This aligns with existing evidence showing PPI efficacy in preventing gastric mucosa damage in patients admitted to ICUs [28, 29]. However for patients with a history of ulcer disease, our results indicate the need for additional preventive measures. It may be crucial to create more tailored prophylactic strategies. Further research is needed in this area, as higher doses of PPI does not necessarily lead to better outcomes in crucial parameters such as mortality and duration of hospitalization [29]. Intermittent therapy is worth consideration as it gives similar results as continuous therapy and is easy to implement. Such therapy is also cheaper and allows better resources utilization. Searching for alternative prophylactic agents may be needed as two most common medications – PPIs and H₂RAs does not show significant differences in patients outcomes [29].

The study took into consideration sizeable population. Despite the presence of other risk factors, such as chronic kidney disease, acute liver injury, NSAIDs use, antiplatelet agents use, or other common demographic variables, none of them showed a significant association with gastric mucosa erosion in our patient group. This however could be due to low power to detect this associations in our study group. As far as NSAIDs use is concerned, the prevalence of gastric mucosa lesion among NSAIDs users is approximately 41% [30]. The prevalence of normal gastric mucosa in NSAIDs users is not well known, however in the general adult population up to 20% use these medications for pain relief. Taking these numbers into consideration, and numbers of cases and controls in our study, the power of our study to detect association between NSAIDs use and gastric mucosa lesion, using alpha level of 0.05, is only 0.12.

Noting that a history of peptic ulcer disease is a significant risk factor, future research could focus on determining whether more personalized pharmacotherapy approaches for a given patient have different effects. It would also be important to expand the criteria for potential risk factors such as genetic predispositions, microbiome composition or inflammatory markers.

Innovative in our study is the way in which we decided to answer a research question. Choosing specifically the group of patients who underwent PEG in the ICU gave us opportunity to retrieve objective data regarding gastric mucosa lesions in a retrospective fashion. Obtaining such data in a prospective fashion would have required much more resources. Our approach allowed us to study a large group of patients in a shorter time.

While our study provides valuable insights, retrospective character and the size of the study group are evident limitations. The size of the study group led to the study being underpowered for important risk factors for gastric mucosa lesion like the use of NSAIDS or antiplatelet agents. The issue we raised should be addressed in further studies with larger number of patients. Prospective cohort studies may be needed to confirm the results of our study. Another limitation is the fact that the results of the study pertain only to critically ill patients in whom PEG was implanted, the results of the study cannot be generalized to all patients in whom PEG was implanted.

Conclusions

Majority of various risk factors for gastric ulceration may not predict gastric mucosa lesions in critically ill patients receiving standard pharmacological prophylaxis undergoing gastroscopy for PEG implantation. History of gastric ulcer may be the only risk factor for gastric ulceration in the ICU patients. Patients with history of gastric ulcer might benefit from higher than standard doses of anti-ulcer medication when hospitalized in the ICU.

Abbreviations

aPTT

Activated partial thromboplastin time

AUROC

Area under receiver operating characteristic curve

BMI

Body mass index

CI

Confidence interval

H₂RA

Histamine type 2 receptor antagonist

ICU

Intensive care unit

IQR

Interquartile range

Me

Median value

NE

Norepinephrine

NG

Nasogastric

OR

Odds ratio

PEG

Percutaneous endoscopic gastrostomy

PPI

Proton pump inhibitor

SD

Standard deviation

Author contributions

P.F.C. – come up with the concept and design for the study, performed analysis, interpreted data, wrote the main manuscript text, prepared figure; K.B., K.D., M.G., J.O. – acquired data, wrote the main manuscript text; P.W. – revised the manuscript and provided important suggestions. All authors reviewed the manuscript.

Funding

Not applicable.

Data availability

The datasets used and/or analyzed during the current study are available from the corresponding author on reasonable request.

Declarations

Ethics approval and consent to participate

Due to retrospective character of the study, the Bioethics Committee of Medical University of Silesia decided that the study does not require an ethical review and waived the requirement for an informed consent (Ethics Committee of Medical University of Silesia in Katowice decision number: PCN/CBN/0022/KB/292/21). The study was conducted according to the principles expressed in the Declaration of Helsinki.

Consent for publication

Not applicable.

Competing interests

The authors declare no competing interests.