Gastric Section Correlation Network for Gastric Precancerous Lesion Diagnosis.

Goal: Diagnosing the corpus-predominant gastritis index (CGI) which is an early precancerous lesion in the stomach has been shown its effectiveness in identifying high gastric cancer risk patients for preventive healthcare. However, invasive biopsies and time-consuming pathological analysis are required for the CGI diagnosis. Methods: We propose a novel gastric section correlation network (GSCNet) for the CGI diagnosis from endoscopic images of three dominant gastric sections, the antrum, body and cardia. The proposed network consists of two dominant modules including the scaling feature fusion module and section correlation module. The front one aims to extract scaling fusion features which can effectively represent the mucosa under variant viewing angles and scale changes for each gastric section. The latter one aims to apply the medical prior knowledge with three section correlation losses to model the correlations of different gastric sections for the CGI diagnosis. Results: The proposed method outperforms competing deep learning methods and achieves high testing accuracy, sensitivity, and specificity of 0.957, 0.938 and 0.962, respectively. Conclusions: The proposed method is the first method to identify high gastric cancer risk patients with CGI from endoscopic images without invasive biopsies and time-consuming pathological analysis.

10-Day versus 14-day bismuth quadruple therapy for first-line eradication of Helicobacter pylori infection: a randomised, open-label, non-inferiority trial.

Background: Bismuth quadruple therapy is currently consensus recommendation for first-line Helicobacter pylori (H. pylori) treatment; however, the optimal duration is unknown. We compared the efficacy of 10-day bismuth quadruple therapy with that of 14-day bismuth quadruple therapy for first-line eradication. Methods: For our multicentre, parallel randomised, open-label, and non-inferiority study, we recruited H. pylori treatment-naïve patients from one medical centre and one teaching hospital in Taiwan. Patients were randomly assigned (1:1) to receive 10-day (PBMT-10) or 14-day (PBMT-14) bismuth quadruple therapy. The primary outcome was the eradication rate as determined by intention-to-treat (ITT) and per-protocol (PP) analyses. The eradication rates between the two groups were compared using a one-sided α value of 0.025 and a non-inferiority margin of 7%. The secondary outcomes were the rate of adverse effects. The trial is registered with ClincialTrials.gov (NCT04527055). Findings: From August 3, 2020 to April 28, 2023, 313 H. pylori treatment-naïve patients (PBMT-10 = 157; PBMT-14 = 156) were enrolled. 35 patients were excluded from PP analyses. The eradication rates (95% CI) for PBMT-10 and PBMT-14 were respectively 92.4% (88.2%-96.5%) and 92.9% (88.9%-97.0%) by ITT analyses, and 97.9% (95.5%-100.0%) and 99.3% (97.8%-100.0%) by PP analyses. The eradication rates for PBMT-10 were non-inferior to those for PBMT-14 (absolute difference [lower boundary of the one-sided 97.5% CI] -0.6% [-6.7%], PNI = 0.020 in ITT analyses, -1.4% [-5.8%], PNI = 0.007 in PP analyses). The rates of overall adverse effects (54.1% versus 57.1%, P = 0.604) were similar between the two groups; nevertheless, the rates of dizziness (18.5% versus 34.0%, P = 0.003) and vomiting (4.5% versus 12.8%, P = 0.008) were lower in PBMT-10 than in PBMT-14. Interpretation: The 10-day bismuth quadruple therapy was non-inferior to the 14-day therapy as a first-line treatment for eradicating H. pylori infection and had no different rates of overall adverse effects, but less serious adverse events in terms of dizziness and vomiting. Funding: The National Science and Technology Council and Ministry of Health and Welfare, Taiwan.

Evolution of the Correa's cascade steps: A long-term endoscopic surveillance among non-ulcer dyspepsia and gastric ulcer after H. pylori eradication.

BACKGROUND: This study is aimed toward investigating the evolution of each Correa's step after Helicobacter pylori eradication in a long-term follow-up and exploring the factors correlated with a high-risk of gastric cancer. METHODS: A total of 1824 H. pylori-infected subjects were enrolled to receive screening endoscopy. Among them, 491 received surveillance endoscopy. The patients were divided into Correa's steps I to VI, from normal to gastric cancer. A group-based trajectory model was used to classify patients as persistent high-risk status or not. RESULTS: The prevalence rates of positive corpus-predominant gastritis index (CGI) were 20%-40% in all age groups and Correa's steps IV-V increased >35% after 50 years based on screening endoscopy. Successful eradication of H. pylori regressed CGI after the 1st year-and-thereafter (P < 0.05) and decreased Correa's step progression (Relative risk 0.66 [95% CI 0.49-0.89], P = 0.01); however, it did not regress OLGA and OLGIM. Not only in steps IV-V, but also in step III, the patients had a risk of developing gastric cancer (11.13-76.41 and 4.61 per 1000 person-years). Age (Hazard ratio 1.012 [1.003-1.020], P = 0.01), OLGA stages ≥ I (2.127 [1.558-2.903], P < 0.001), and OLGIM stages ≥ I (1.409 [1.119-1.774], P = 0.004) were correlated independently with a persistent high-risk status. CONCLUSION: The patients in Correa's steps III-V, but not I-II, were at risk of gastric cancer after H. pylori eradication. Age, OLGA stages ≥ I, and OLGIM stages ≥ I were independent factors correlated to a persistent high-risk of gastric cancer. The data may be useful when scheduling surveillance endoscopy for subjects in each Correa's step (NCT04527055).

Deep Ensemble Feature Network for Gastric Section Classification.

In this paper, we propose a novel deep ensemble feature (DEF) network to classify gastric sections from endoscopic images. Different from recent deep ensemble learning methods, which need to train deep features and classifiers individually to obtain fused classification results, the proposed method can simultaneously learn the deep ensemble feature from arbitrary number of convolutional neural networks (CNNs) and the decision classifier in an end-to-end trainable manner. It comprises two sub networks, the ensemble feature network and the decision network. The former sub network learns the deep ensemble feature from multiple CNNs to represent endoscopic images. The latter sub network learns to obtain the classification labels by using the deep ensemble feature. Both sub networks are optimized based on the proposed ensemble feature loss and the decision loss which guide the learning of deep features and decisions. As shown in the experimental results, the proposed method outperforms the state-of-the-art deep learning, ensemble learning, and deep ensemble learning methods.

Synthesis and Characterization of CuIn1-xGaxSe2 Semiconductor Nanocrystals.

In this paper, the synthesis and characterization of CuIn1-xGaxSe2 (0 £ x £ 1) nanocrystals are reported with the influences of x value on the structural, morphological, and optical properties of the nanocrystals. The X-ray diffraction (XRD) results showed that the nanocrystals were of chalcopyrite structure with particle size in the range of 11.5-17.4 nm. Their lattice constants decreased with increasing Ga content. Thus, the x value of the CuIn1-xGaxSe2 nanocrystals was estimated by Vegard's law. Transmission electron microscopy (TEM) analysis revealed that the average particle size of the nanocrystals agreed with the results of XRD. Well-defined lattice fringes were shown in the TEM images. An analysis of the absorption spectra indicated that the band gap energy of these CuIn1-xGaxSe2 nanocrystals was tuned from 1.11 to 1.72 eV by varying the x value from 0 to 1. The Raman spectra indicated that the A1 optical vibrational mode of the nanocrystals gradually shifted to higher wavenumber with increasing x value. A simple theoretical equation for the A1 mode frequency was proposed. The plot of this equation showed the same trend as the experimental data.

Spasmolytic polypeptide-expressing metaplasia associated with higher expressions of miR-21, 155, and 223 can be regressed by Helicobacter pylori eradication in the gastric cancer familial relatives.

BACKGROUND AND AIMS: Spasmolytic polypeptide-expressing metaplasia (SPEM) is a preneoplastic gastric cancer lesion related to epigenetic microRNA (miRNA) expression. This study elucidated whether Helicobacter pylori-infected first-degree relatives of patients with gastric cancer (GCF) are susceptible to have SPEM and correlated with miR-21, 155, and 223 expressions. We also validated whether SPEM and these miRNAs can be regressed after H pylori eradication. METHODS: We prospectively enrolled 148 GCF and 148 nonulcer dyspepsia (NUD) subjects without gastric cancer familial history as controls. Each case had received a panendoscopy to determine H pylori status and gastric histology, including SPEM. The cases with SPEM were followed after H pylori eradication to determine SPEM regression. The total RNA was extracted to analyze tissues miR-21, 155, and 223 before and after eradication. RESULTS: GCF subjects had a higher prevalence of H pylori infection (73% vs 32%) and SPEM (42% vs 14%, P < 0.01) than controls. The tissue miR-21, 155, and 223 in antrum were higher in cases with SPEM than in those without SPEM (P <= 0.05). There was similar SPEM reversibility after H pylori eradication between GCF subjects and controls (72% vs 69%, P = 0.852). In the SPEM regressed cases, tissue miR-21, 155, and 223 decreased after H pylori eradication (P < 0.05). CONCLUSION: The H pylori-infected GCF subjects were prone to have SPEM with higher tissues miR-21, 155, and 223 expressions. H pylori eradication can result in a 70% SPEM regression, accompanied by a decline in miR-21, 155, and 233 expression levels.

The corpus-predominant gastritis index can be an early and reversible marker to identify the gastric cancer risk of Helicobacter pylori-infected nonulcer dyspepsia.

BACKGROUND: Corpus-predominant gastritis index (CGI) is an early histological marker to identify Helicobacter pylori-infected gastric cancer relatives at risk of cancer. This study validated whether CGI is more prevalent in H. pylori-infected nonulcer dyspepsia (NUD) subjects than in duodenal ulcer (DU) controls and whether it is reversible after H. pylori eradication or is correlated with noninvasive biomarkers. MATERIALS AND METHODS: In this longitudinal cohort study, 573 H. pylori-infected subjects were enrolled, including 349 NUD and 224 DU. Gastric specimens were provided to assess CGI, spasmolyic polypeptide-expressing metaplasia (SPEM), and Operative Link on Gastric Intestinal Metaplasia assessment (OLGIM). Serum pepsinogen I and II levels were assessed using enzyme-linked immunosorbent assay. CGI subjected were followed up at least 1 year after H. pylori eradication. RESULTS: NUD subjects had higher prevalence rates of CGI (47.0% vs 29.9%, P<.001) and OLGIM stages III-IV (24.1% vs 15.2%, P=.01) than controls. CGI was highly prevalent in NUD subjects after the age of 40, which was 10 years earlier than atrophic gastritis and intestinal metaplasia. NUD subjects with CGI had higher risk of SPEM (OR 2.86, P<.001) and lower serum pepsinogen I/II ratios (P<.001) than those without CGI. Serum pepsinogen I/II ratios <9 could predict CGI modestly (AUROC 0.69, 95% CI: 0.63-0.74). CGI was regressed after eradication (P<.001). CONCLUSIONS: CGI was more prevalent in H. pylori-infected NUD subjects than in controls, was correlated with SPEM, and may serve as a marker earlier than OLGIM to indicate risk of gastric cancer. Moreover, CGI could be regressed after eradication.

Serum Level of Trefoil Factor 2 can Predict the Extent of Gastric Spasmolytic Polypeptide-Expressing Metaplasia in the H. pylori-Infected Gastric Cancer Relatives.

BACKGROUND & AIMS: Gastric cancer has familial clustering in incidence, and the familial relatives of gastric cancer sufferers are prone to have spasmolytic polypeptide-expressing metaplasia (SPEM), and intestinal metaplasia (IM) after H. pylori infection. This study tested whether serum pepsinogen I/II and trefoil factor family (TFF) proteins can predict SPEM or IM in the H. pylori-infected relatives of patients with gastric cancer. METHODS: We prospectively enrolled 119 H. pylori-infected relatives of gastric cancer patients of noncardiac gastric cancer patients, who then received panendoscopy to obtain gastric biopsy to define the presence of corpus gastritis index (CGI), SPEM, and IM. The advanced SPEM in histology was defined by TFF2 immunohistochemistry. Each patient also had checkups of serum TFF2, TFF3, and pepsinogen I/II by enzyme-linked immunosorbent assay (ELISA). RESULTS: The 119 H. pylori-infected relatives included 61 with SPEM, and 34 with IM. The presence of either IM or SPEM was not related to the serum TFF2, TFF3, and pepsinogen I/II levels (p > .05). Serum TFF2 levels were higher in relatives with CGI who also had advanced SPEM (p = .032). For relatives without CGI, the elevated serum TFF2 levels correlated with higher H. pylori density and more severe gastritis in antrum (p = .001). CONCLUSION: The serum TFF2 level cannot predict SPEM or IM in H. pylori-infected relatives of patients with gastric cancer. For H. pylori-infected relatives with CGI, serum TFF2 levels may predict the advanced severity of SPEM. Elevated serum TFF2 levels may indicate severe H. pylori-related inflammation, at risk of development or progression of SPEM in relatives without CGI.

Genomic single nucleotide polymorphisms in the offspring of gastric cancer patients predispose to spasmolytic polypeptide-expressing metaplasia after H. pylori infection.

BACKGROUND: Gastric cancer exhibits familial clustering, and gastric cancer familial relatives (GCF) tend to present with corpus-predominant gastritis and precancerous lesions as SPEM or IM after H. pylori infection. The study determined whether the children of gastric cancer patients (GCA) had genomic single nucleotide polymorphisms (SNPs) predisposed to the gastric precancerous lesions as spasmolytic polypeptide-expressing metaplasia (SPEM) or intestinal metaplasia (IM). RESULTS: There were 389 family relatives of 193 non-cardiac GCA and 173 duodenal ulcer patients (DU), received blood sampling for DNA collection. The differences of the risk alleles of SNPs in the ITGA5, ITGB1, IL-10, COX-2, RUNX3, and TFF2 genes were compared between 195 children of GCA and 143 DU. The children of GCA had higher allele frequencies of ITGA5-1160 T-carrier (P = 0.006, OR[95% CI] = 2.2[1.2-4]), ITGB1-1949 A-carrier (P = 0.047; OR[95% CI] = 2.8[1.4-5.3]), ITGB1 + 31804 C-carrier (P = 0.013; OR[95% CI] = 4.7[1.7-13.0]), IL-10-592 AA (P = 0.014; OR[95% CI] = 2.3[1.4-4.0]) and COX-2-1195 G-carrier (P = 0.019; OR[95% CI] = 1.7[0.9-3.2]) than DU. The combined genotype with ITGA5-1160/ITGB1-1949/ITGB1 + 31804 as T/A/C carriers and COX-2-1195/IL-10-592 as G-carrier/AA was more prevalent in the children of GCA than in DU (P < 1×10(-4)), and predisposed with a 5.3-fold risk of getting SPEM in the H. pylori-infected children of GCA (P = 0.016). Such risk of getting SPEM increased to 112 folds, if combined with RUNX3 + 492/TFF2-308 as A-carrier/CC in this limited study scale (P = 1×10(-4)). CONCLUSIONS: The SNPs of ITGA5-1160/ITGB1-1949/ ITGB1 + 31804 as T/A/C carriers and COX-2-1195/IL-10-592 as G-carrier/AA, or more specific to combine RUNX3 + 492/TFF2-308 as A-carrier/CC shall be host factor predisposing to gastric cancer during H. pylori infection, and serve as marker to identify high-risk subjects for H. pylori eradication.

Intravenous albumin shortens the duration of hospitalization for patients with hypoalbuminemia and bleeding peptic ulcers: a pilot study.

BACKGROUND: Patients with hypoalbuminemia have an increased risk of ulcer rebleeding and longer length of hospitalization. AIMS: This study aimed to test whether intravenous albumin can decrease the incidence of rebleeding or shorten the duration of hospitalization in patients with bleeding peptic ulcers and hypoalbuminemia. METHODS: Sixty-two patients with bleeding peptic ulcers and Rockall scores ≥ 6 were prospectively enrolled after having received endoscopic therapy. The enrolled patients were divided into a normal albumin group (serum albumin ≥ 3 g/dL, n = 39) or an intervention group (<3 g/dL, n = 23) to receive a 3-day course of omeprazole infusion and 25-day oral esomeprazole. Patients (n = 29) with bleeding ulcers and hypoalbuminemia who received the same dose of intravenous and oral omeprazole but did not receive albumin therapy were enrolled from a previous study as the control group. In the intervention group, patients received albumin infusion (10 g q8h) for 1 day (serum albumin levels 2.5-2.9 g/dL) and 2 days (<2.5 g/dL), respectively. RESULTS: The 28-day cumulative rebleeding rates were similar between the intervention group and the control group (39.1 vs. 42.3 %, p = 0.99). The intervention group had a shorter duration of hospitalization (9 vs. 15 days, p = 0.02) than cohort controls. The risk of rebleeding developed after discharge were similar (normal albumin group vs. intervention group vs. control group, 1/5 [20 %] vs. 2/9 [22.2 %] vs. 1/11 [9.1 %], p = 0.7). CONCLUSIONS: Albumin administration shortens the duration of hospitalization for patients with peptic ulcer bleeding and hypoalbuminemia, but does not decrease the incidence of rebleeding.

Long-term celecoxib can prevent the progression of persistent gastric intestinal metaplasia After H. pylori eradication.

BACKGROUND AND AIM: Intestinal metaplasia (IM) has overexpressions of COX-2. Short-term 8-week celecoxib, a selective COX-2 inhibitor, exerts a preliminary hint to improve regression in part for persistent IM after Helicobacter pylori eradication. This study further validated whether or not a prolonged duration of celecoxib of up to 1 year can be safe and effective. METHODS: One hundred and forty patients, with persistent IM after H. pylori eradication for 1 year, were included with half of them receiving celecoxib 200 mg/day for 12 months and the other half serving as controls. Each patient received serial checkups of blood creatinine levels every 4 months. After the 1-year follow-up, panendoscopy was repeated to assess the IM regression. The serial gastric specimens, taken before and after celecoxib therapy, were immunochemically stained for COX-2. RESULTS: The intention-to-treat (ITT) and per-protocol (PP) analyses to the rates of IM regression were higher in the celecoxib group than in the controls (ITT: 44.3% [31/70] vs 14.3% [10/70], p < .001; and PP: 51.7% [31/60] vs 16.1% [10/62], p < .001). All enrolled patients had no renal impairment during follow-up. Even in the patients without IM regression, the mean IM scores and COX-2 expressions were significantly more decreased in the celecoxib group than in the controls (p < .005). CONCLUSION: One year 200-mg celecoxib daily be safely administered to improve the regression or prevent the progression of persistent IM after H. pylori eradication.

Weak up-regulation of serum response factor in gastric ulcers in patients with co-morbidities is associated with increased risk of recurrent bleeding.

BACKGROUND: Serum response factor (SRF) is crucial for gastric ulcer healing process. The study determined if gastric ulcer tissues up-regulate SRF and if such up-regulation correlated with co-morbidities and the risk of recurrent bleeding. METHODS: Ulcer and non-ulcer tissues were obtained from 142 patients with active gastric ulcers for SRF expression assessed by immunohistochemistry. Based on the degree of SRF expression between these two tissue types, SRF up-regulation was classified as strong, intermediate, and weak patterns. The patients were followed-up to determine if SRF up-regulation correlated to recurrent bleeding. RESULTS: Gastric ulcer tissues had higher SRF expression than non-ulcer tissues (p < 0.05). Patients with strong SRF up-regulation had lower rates of stigmata of recent hemorrhage (SRH) on the ulcer base than the others (p < 0.05). Multivariate logistic regression confirmed that co-morbidities and weak SRF up-regulation were two independent factors of recurrent gastric ulcer bleeding (p < 0.05). Combining both factors, there was an 8.29-fold (95% CI, 1.31~52.62; p = 0.03) higher risk of recurrent gastric ulcer bleeding. CONCLUSIONS: SRF expression is higher in gastric ulcer tissues than in non-ulcer tissues. Weak SRF up-regulation, combined with the presence of co-morbidities, increase the risk of the recurrent gastric ulcer bleeding.

Supplementation of Los Angeles classification with esophageal mucosa index of hemoglobin can predict the treatment response of erosive reflux esophagitis.

BACKGROUND: We assessed whether the esophageal mucosa index of hemoglobin (IHb) could assist the Los Angeles (LA) classification in defining the severity of erosive reflux esophagitis (RE) and predicting the treatment response by esomeprazole. METHODS: Five hundred twenty-four subjects (424 RE and 100 controls) with normal body mass index and hemoglobin had undergone endoscopy to confirm the RE grade by LA classification and to check the series of IHb values at every centimeter of the esophageal mucosa while withdrawing the endoscope to above the esophageal-gastric junction (EGJ). The RE cases had received esomeprazole for 8 weeks (40 mg/day) to assess the cumulative proportions of sustained symptomatic response (SSR). RESULTS: The IHb value at the EGJ was higher in RE patients than in controls (P < 0.001). Selecting 70 as the cutoff IHb value at the EGJ, the efficacy to define RE achieved 95.8% (406/424) sensitivity and 94% (94/100) specificity. For the patients with same the LA grade, the length of IHb value >70 above the EGJ (L-IHb70) that is greater than 4 cm correlated with a poor cumulative rate of SSR (P < 0.01). CONCLUSION: The IHb value >70 at the EGJ is a reliable indicator of RE. The L-IHb70 can serve as a supplementary indicator to predict the response to esomeprazole index within the same LA grade.

Double-dosed pantoprazole accelerates the sustained symptomatic response in overweight and obese patients with reflux esophagitis in Los Angeles grades A and B.

OBJECTIVES: Body mass index (BMI) in the range defined as overweight or obese adversely decreases the sustained symptomatic response (SSR) to proton pump inhibitors for patients with reflux esophagitis of Los Angeles grade A or B (RE-AB). We thus investigated whether double-dosed pantoprazole can accelerate SSR in such patients. METHODS: A total of 200 overweight or obese patients with RE-AB were evenly randomized into a double-dosed group (receiving 8-week pantoprazole 40 mg twice daily) or a standard-dosed control group (receiving 8-week pantoprazole 40 mg per day and one blank tablet at night). In each patient, demographic factors and the genotype of S-mephenytoin 4'-hydroxylase (CYP2C19) were checked and defined as poor metabolizer (PM), or homologous extensive metabolizer (HomoEM), or heterologous extensive metabolizer (HeteroEM). The cumulative proportions of patients with SSR were compared during the 8-week period. RESULTS: Both intention-to-treat and per-protocol analyses disclosed that the rates of SSR were higher in the double-dosed group than in the standard-dosed group from week 4 (P=0.005) until week 8 (P=0.01). While using standard-dosed pantoprazole, PMs had better rates of SSR during the 8-week period than both HomoEMs and HeteroEMs (P<0.05). By using double-dosed pantoprazole, the cumulative rates of SSR were improved as early as week 4 for both HomoEMs and HeteroEMs (P<0.005, log-rank test). CONCLUSIONS: For RE-AB in overweight and obese patients, double-dosed pantoprazole effectively accelerates the SSR, especially for those with CYP2C19 genotypes as HeteroEM or HomoEM. Accordingly, it offers an earlier shift into on-demand pantoprazole for RE-AB patients with high BMI.

Seven-day intravenous low-dose omeprazole infusion reduces peptic ulcer rebleeding for patients with comorbidities.

BACKGROUND: Patients with comorbidities have an increased risk of ulcer rebleeding, especially within the 28 days after endoscopic therapy. Omeprazole infusion can prevent rebleeding after endoscopic therapy in patients with peptic ulcer bleeding. However, the optimal duration is uncertain, especially for those patients with comorbidities. OBJECTIVE: To determine whether prolonged low-dose intravenous omeprazole could reduce rebleeding for patients with comorbidities. DESIGN: A prospective randomized control study. SETTING: National Cheng Kung University, Tainan, Taiwan. PATIENTS: A total of 147 patients with comorbidities and peptic ulcer bleeding controlled by endoscopic hemostasis were enrolled. INTERVENTIONS: The enrolled patients were randomized into either the 7-day low-dose group or the 3-day high-dose group, who received 3.3 mg/h or 8 mg/h continuous omeprazole infusion, respectively. After omeprazole infusion, oral esomeprazole 40 mg every day was given. MAIN OUTCOME MEASUREMENTS: To compare the rebleeding rates within 28 days after gastroscopy between the 2 study groups. RESULTS: The 7-day cumulative rebleeding rate was similar between the 2 groups (9.5% vs 9.7%, P > .05), but the 7-day low-dose group had a lower risk of rebleeding between the 8th and 28th day compared with the 3-day high-dose group (0% vs 10.7%, P = .03; relative risk, 0.52 [95% CI, 0.43-0.63]). The Kaplan-Meier curves confirmed that the 7-day low-dose group had a significantly higher cumulative rebleeding-free proportion between the 8th and 28th day than the 3-day high-dose group (P = .02, log-rank test). CONCLUSIONS: In Asian patients, prolonged low-dose omeprazole infusion for 7 days may reduce peptic ulcer rebleeding during the first 28 days in patients with comorbidities.