Objective The association between the severity of COVID-19 and gastrointestinal (GI) bleeding is unknown. This study aimed to determine whether the severity of COVID-19 is a risk factor for GI bleeding.
Design A multicentre, retrospective cohort study was conducted on hospitalised patients with COVID-19 between January 2020 and December 2021. The severity of COVID-19 was classified according to the National Institute of Health severity classification. The primary outcome was the occurrence of GI bleeding during hospitalisation. The main analysis compared the relationship between the severity of COVID-19 and the occurrence of GI bleeding. Multivariable logistic regression analysis was performed to evaluate the association between the severity of COVID-19 and the occurrence of GI bleeding.
Results 12 044 patients were included. 4165 (34.6%) and 1257 (10.4%) patients had severe and critical COVID-19, respectively, and 55 (0.5%) experienced GI bleeding. Multivariable analysis showed that patients with severe COVID-19 had a significantly higher risk of GI bleeding than patients with non-severe COVID-19 (OR: 3.013, 95% CI: 1.222 to 7.427). Patients with critical COVID-19 also had a significantly higher risk of GI bleeding (OR: 15.632, 95% CI: 6.581 to 37.130). Patients with severe COVID-19 had a significantly increased risk of lower GI bleeding (OR: 10.349, 95% CI: 1.253 to 85.463), but the risk of upper GI bleeding was unchanged (OR: 1.875, 95% CI: 0.658 to 5.342).
Conclusion The severity of COVID-19 is associated with GI bleeding, and especially lower GI bleeding was associated with the severity of COVID-19. Patients with severe or critical COVID-19 should be treated with caution as they are at higher risk for GI bleeding.
- GASTROINTESTINAL BLEEDING
Data availability statement
Data are available upon reasonable request.
This is an open access article distributed in accordance with the Creative Commons Attribution Non Commercial (CC BY-NC 4.0) license, which permits others to distribute, remix, adapt, build upon this work non-commercially, and license their derivative works on different terms, provided the original work is properly cited, appropriate credit is given, any changes made indicated, and the use is non-commercial. See: http://creativecommons.org/licenses/by-nc/4.0/.
Statistics from Altmetric.com
If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.
WHAT IS ALREADY KNOWN ON THIS TOPIC
Previous studies reported an association between COVID-19 and gastrointestinal (GI) bleeding, but the association between the severity of COVID-19 and GI bleeding is unknown.
WHAT THIS STUDY ADDS
Patients with severe or critical COVID-19 had a significantly higher risk of GI bleeding than non-severe patients. Patients with severe COVID-19 had a significantly increased risk of lower GI bleeding, but the risk of upper GI bleeding was unchanged.
HOW THIS STUDY MIGHT AFFECT RESEARCH, PRACTICE OR POLICY
GI bleeding, especially lower GI bleeding, was associated with the severity of COVID-19. Clinicians should be aware that patients with severe COVID-19 are at higher risk of GI bleeding.
The COVID-19 progressed to become a global pandemic since its first case in China in 2019. In Japan, 1 728 228 cases have been confirmed by the end of 2021.1 COVID-19 reportedly causes thrombotic complications due to vascular damage caused by SARS-CoV-2,2 3 including in the gastrointestinal (GI) tract. Several observational studies reported an association between COVID-19 and GI bleeding.4–7 A European study using registry data of 4128 patients reported a GI bleeding rate of 1.6% in hospitalised patients with COVID-19 and that anticoagulation therapy was associated with a significantly increased risk of GI bleeding. Univariate analysis also showed a correlation between the severity of COVID-19 and GI bleeding.6 An analysis of 1158 hospitalised patients with COVID-19 in the USA reported a GI bleeding rate of 3% during hospitalisation; 70% of the cases of GI bleeding involved the upper GI tract.7
Previous reports have only analysed the association between factors in hospitalised patients with COVID-19 and GI bleeding, and only one study analysed the association between the severity of COVID-19 and GI bleeding.6 Therefore, it is unclear whether the severity of COVID-19 is associated with GI bleeding. Identifying risk factors for GI bleeding based on the severity of COVID-19 will provide useful information for real-world clinical situations, such as managing the bleeding risk according to the severity of COVID-19 in hospitalised patients and determining the indications for emergency haemostatic procedures. The prognosis of COVID-19 reportedly improves following anticoagulation for thrombosis.8–12 Since anticoagulants increase the risk of GI bleeding, it is important to identify patients with COVID-19 at high risk of GI bleeding.
In this study, we collected and examined datasets of hospitalised patients with COVID-19 from multiple hospitals in and around Tokyo, Japan, to determine whether the severity of COVID-19 is a risk factor of GI bleeding.
Materials and methods
We used an electronic medical record system to identify patients admitted due to COVID-19 confirmed by PCR in the participating study hospitals between January 2020 and December 2021. Outpatients were not included in the study. Cumulatively, we identified 12 206 patients.
According to data from the Ministry of Health, Labour and Welfare of Japan, the total number of individuals who have contracted COVID-19 in Japan is 1 307 341 as of 31 December 20211; hence, 0.93% of patients who contracted COVID-19 in Japan were included in this study.
Patients with signs of GI bleeding on admission were excluded because of GI bleeding possibly occurring prior to infection with COVID-19 (n=16). Patients with missing information were also excluded from the study (n=146). Finally, 12 044 patients were included in the final analysis (online supplemental file 1).
All medical records were anonymised before performing statistical analyses. We displayed information regarding the study in participating hospitals and obtained informed consent in the form of opt-out. The study was conducted in accordance with the Declaration of Helsinki.
The primary endpoint was the occurrence of GI bleeding during hospitalisation. We collected background information of the patients, including gender, age, date of admission, medical history, and medications taken at admission. The medical history was evaluated for known risks of GI bleeding, such as history of GI bleeding, chronic kidney disease (CKD),13 liver cirrhosis, history of cancer,14 and inflammatory bowel disease. We also assessed whether patients were taking aspirin, antiplatelet agents other than aspirin, warfarin, novel oral anticoagulants (NOAC), non-steroidal anti-inflammatory drugs (NSAIDs), corticosteroids, proton-pump inhibitors (PPI)/potassium-competitive acid blocker (P-cab), and H2 receptor blockers (H2RB), all of which are associated with GI bleeding.15
The severity of COVID-19 was assessed at the most severe point during hospitalisation. In this study, severity was classified according to the National Institute of Health classification,16 which has five levels: asymptomatic or presymptomatic infection, mild, moderate, severe, and critical. Cases not requiring oxygen supplementation were classified as non-severe in this study; otherwise, they were classified as severe or critical. Since respiratory failure was defined as critical COVID-19, cases requiring mechanical ventilation were classified as critical. Therefore, patients were stratified into three levels in this study: non-severe, severe, and critical COVID-19.
The presence of GI bleeding during hospitalisation was determined from the medical records. Patients with signs of GI bleeding (haematemesis, bloody stools, bleeding from an indwelling GI tube, or melena) and a confirmed diagnosis of GI bleeding by diagnostic imaging tests (CT or endoscopy) were considered to have GI bleeding. Information was collected on the diagnosis of the cause of bleeding as well as whether it was upper or lower GI bleeding.
In this study, it was also necessary to determine the risk of significant GI bleeding as its occurrence may require emergency endoscopic treatment.15 Therefore, we conducted a separate study to examine the association between the severity of COVID-19 and significant GI bleeding. Significant GI bleeding was defined as a decrease of 20 g/L or more in haemoglobin level at the time of GI bleeding from the baseline haemoglobin level on admission.
We also evaluated the treatment during hospitalisation according to the severity of COVID-19 to determine the association between posthospitalisation treatment and GI bleeding.
In addition, we investigated whether there was a difference in the pathogenesis of GI haemorrhage between COVID-19 and non-COVID-19 pneumonia. Using an electronic medical record system, we collected the causes of GI bleeding among pneumonia admissions other than COVID-19 with GI bleeding at Tokyo Medical and Dental University between January 2020 and December 2021.
To compare baseline characteristics between the three COVID-19 severity classifications and background factors, the χ2 test or Fisher’s exact test was used for categorical variables, and the Kruskal-Wallis test was used for continuous variables.
The primary analysis assessed the relationship between the severity of COVID-19 and GI bleeding by adopting logistic regression analysis models. First, confounding factors were examined in the univariate analysis, and significant confounders (p<0.05) were included in the multivariate analysis, which was adjusted for covariates. To adjust for the potential difference between healthcare institutions, a model adjusting for the inpatient hospital as a confounding factor was also examined.
As subanalysis, we performed logistic regression analysis to examine (1) the association between significant GI bleeding and the severity of COVID-19, (2) the association between upper GI bleeding and the severity of COVID-19, and (3) the association between lower GI bleeding and the severity of COVID-19.
Additionally, we evaluated the association between posthospitalisation treatment and GI bleeding according to the severity of COVID-19 using descriptive statistics. The χ2 test or Fisher’s exact test was used for categorical variables.
All analyses were performed using Stata/MP V.16.1 (StataCorp LLC, College Station, Texas, USA). A two-sided p value of <0.05 was considered statistically significant.
Table 1 shows the patient background and the results of the primary endpoint. Overall, 12 044 patients were included in the study. According to the severity of COVID-19, there were 6622 (55.0%), 4165 (34.6%), and 1257 (10.4%) patients with non-severe, severe, and critical COVID-19. The median age of patients was 47, 57, and 67 years in the non-severe, severe, and critical COVID-19 groups, with increasing age correlating to an increased severity of the disease (p<0.001). There was also a significantly greater proportion of men in the critical COVID-19 group (p<0.001). The use of aspirin, other antiplatelet agents, warfarin, NOACs and corticosteroids was significantly more frequent as the severity of the disease worsened. The proportion of patients taking PPI/P-cabs and H2RBs in the severe COVID-19 group was significantly higher than the other groups. There was a significant association between the date of hospitalisation and COVID-19 severity. Online supplemental file 2 shows the severity of COVID-19 according to the healthcare institution. The distribution of the severity of COVID-19 was significantly different among inpatient hospitals (p<0.001).
Result of GI bleeding
Table 2 shows the results of the primary endpoint. GI bleeding occurred in 55 patients (0.5%), with significantly more patients experiencing GI bleeding as the disease worsened (p<0.001). Of the 55 patients with GI bleeding, 38 (69.1%) had upper GI bleeding, and 18 (32.7%) had lower GI bleeding; one patient had both upper and lower GI bleeding. A significant association was found between upper and lower GI bleeding and the severity of COVID-19 (both p<0.001). Online supplemental file 3 shows the demographics of patients with GI bleeding. The median decrease in haemoglobin level in GI bleeding cases was 22 g/L, with no association between the amount of decrease in haemoglobin level and severity of COVID-19.
Online supplemental file 4 shows the demographics of patients with significant GI bleeding. Overall, 35 patients met the criteria for significant GI bleeding. There was a significant association between COVID-19 severity and significant GI bleeding (p<0.001).
Online supplemental file 5 shows the results of diagnostic imaging tests performed to evaluate upper GI bleeding. Acute gastric mucosal lesions (AGML) and gastric ulcers were the most common causes of upper GI bleeding (n=9). Online supplemental file 6 shows the results of tests performed to evaluate lower GI bleeding. Diverticular bleeding and rectal ulcers were the most common causes of lower GI bleeding (n=5). There were several pathological conditions suggestive of impaired blood flow, such as colorectal and rectal ulcers, ischaemic enterocolitis, and necrotising enterocolitis.
GI bleeding is associated with the severity of COVID-19
A logistic regression analysis model was adopted to examine risk factors for GI bleeding in hospitalised patients with COVID-19 (table 3). First, univariate analysis was performed to examine the association of each confounding factor with the risk of GI bleeding. Male sex, older age, history of GI bleeding, CKD, cancer, use of aspirin, NOACs and PPI/P-cabs were significant risk factors for GI bleeding. Regarding the timing of hospitalisation, there was no significant difference in the bleeding risk depending on the date of hospitalisation. Regarding the severity of COVID-19, patients with severe (OR: 3.644, 95% CI: 1.498 to 8.866) and critical (OR: 24.686, 95% CI: 10.871 to 56.054) COVID-19 had a significantly increased risk of GI bleeding.
We then performed multivariate analysis after adjusting for significant factors found in the univariate analysis. There was a significantly increased risk of GI bleeding in patients with severe (OR: 3.013, 95% CI: 1.222 to 7.427) and critical (OR: 15.632, 95% CI: 6.581 to 37.130) COVID-19. Additionally, male sex (OR: 0.482, 95% CI: 0.236 to 0.985), history of GI bleeding (OR: 9.818, 95% CI: 4.030 to 23.920) and NOAC use (OR: 4.394, 95% CI: 1.833 to 10.533) were significant risk factors for GI bleeding.
To determine bias according to the healthcare institution, additional analysis was performed that included the inpatient hospital as a confounding factor (online supplemental file 7). The association between the severity of COVID-19 and GI bleeding was significant after adjusting for difference in the admitted hospitals. There were no significant differences in the risk of GI bleeding by healthcare institution.
Association between the severity of COVID-19 and significant GI bleeding
Logistic regression analysis with significant GI bleeding as the outcome was performed to examine its risk factors in admitted patients with COVID-19 (online supplemental file 8). After adjusting for covariates, multivariate analysis revealed that patients with severe (OR: 3.396, 95% CI: 1.068 to 10.799) and critical (OR: 15.532, 95% CI: 5.059 to 47.687) COVID-19 had an increased risk of severe GI bleeding. Additionally, a history of GI bleeding and use of NOACs were significantly associated with an increased risk of severe GI bleeding.
Severe COVID-19 increases the risk of lower GI bleeding
Multivariate analysis was performed to determine the risk of upper and lower GI bleeding according to the severity of COVID-19.
Table 4 shows the results of logistic regression analysis with upper GI bleeding as the outcome. After adjusting for covariates, patients with critical COVID-19 had a significantly increased risk of upper GI bleeding (OR: 12.114, 95% CI: 4.663 to 31.473), while the risk was unchanged in patients with severe COVID-19 (OR: 1.875, 95% CI: 0.658 to 5.342). The results for upper GI bleeding were different from those for overall GI bleeding, as the risk of upper GI bleeding was not increased in patients with severe COVID-19.
Table 5 shows the results of logistic regression analysis with lower GI bleeding as the outcome. After adjusting for covariates, patients with severe (OR: 10.349, 95% CI: 1.253 to 85.463) and critical (OR: 37.298, 95% CI: 4.527 to 307.292) COVID-19 had a significantly increased risk of lower GI bleeding. These results revealed that severe COVID-19 is a risk factor for lower GI bleeding, but not for upper GI bleeding.
Association between GI bleeding and COVID-19 treatment after hospitalisation
We assessed the details of COVID-19 treatment given during hospitalisation according to COVID-19 severity to evaluate the association between treatment and GI bleeding. Online supplemental file 9 shows the treatment of COVID-19 according to its severity. Both antiviral and anticoagulant medications were used increasingly as the severity of COVID-19 increased.
Online supplemental files 10–12 show the association between GI bleeding and treatment according to the severity of COVID-19. In the non-severe COVID-19 group (online supplemental file 10), PPI/P-cab was used in six of the seven cases of GI bleeding (p<0.001). There was no significant association between anticoagulation and GI bleeding. In the severe COVID-19 group (online supplemental file 11), continuous haemodiafiltration (CHDF) was administered in 10 patients, but no cases of GI bleeding were observed. There was also no significant association between anticoagulation and GI bleeding in this group. Among patients with critical COVID-19 (online supplemental file 12), GI bleeding more frequently in those who received NOACs (p=0.019), CHDF (p=0.009), extracorporeal membrane oxygenation (ECMO) (p=0.001), and PPI/P-cabs (p<0.001).
Multivariate analysis revealed that severe COVID-19 was a risk factor for lower GI bleeding (table 5), but anticoagulant treatment was not (online supplemental file 11). Among patients with critical COVID-19, GI bleeding occurred more frequently in patients using potent anticoagulants, such as those who underwent CHDF and ECMO (online supplemental file 12). The higher risk of GI bleeding in patients with critical COVID-19 may be due to posthospitalisation therapy given for COVID-19.
Comparison of the pathogenesis of GI bleeding associated with COVID-19 pneumonia and non-COVID-19 pneumonia
Lastly, we examined whether the pathogenesis of GI bleeding differed between COVID-19 pneumonia and non-COVID-19 pneumonia. Of the 1173 cases hospitalised with pneumonia other than COVID-19, 27 cases of non-COVID-19 pneumonia (2.3%) were complicated by GI bleeding, 13 of which had upper GI bleeding (1.1%) and 14 of which had lower GI bleeding (1.2%).
Online supplemental file 13 shows the results of diagnostic imaging tests performed to evaluate GI bleeding in non-COVID-19 pneumonia cases. Of the 13 cases of upper GI bleeding, 5 (38%) had duodenal ulcer and 4 (31%) had gastric ulcer as the cause of bleeding. The main cause of upper GI bleeding was ulcerative disease.
Of the 14 cases of lower GI bleeding, diverticular bleeding (n=7) was the most common cause. Conditions suggestive of impaired blood flow were present in 4 cases (28%), including ischaemic enterocolitis (n=1) and rectal ulcer (n=3). The rate of conditions suggestive of reduced blood flow in non-COVID-19 pneumonia was lower than that in COVID-19 pneumonia (online supplemental file 6, 39% in 7 out of 18 cases of lower GI bleeding: 1 ischaemic enterocolitis, 1 necrotising enterocolitis, and 5 rectal ulcers).
This study clarified the association between COVID-19 severity and GI bleeding using real-world data from 12 044 hospitalised patients with COVID-19. The results showed that GI bleeding, whether severe or non-severe, and especially lower GI bleeding was associated with the severity of COVID-19. Interestingly, this study also revealed that severe COVID-19 was a risk factor for lower GI bleeding but not for upper GI bleeding. To the best of our knowledge, this is the first study to collect and analyse medical records of more than 10 000 patients to determine the association between GI bleeding and COVID-19 severity.
In this study, GI bleeding occurred in 0.5% of hospitalised patients with COVID-19. In a previous report from Europe, the proportion of hospitalised patients with COVID-19 in whom GI bleeding occurred was 0.9%.6 The probability of GI bleeding in the previous report was a slightly larger percentage than in our cohort. The percentage in this study may have been smaller than in the previous report because the definition of GI bleeding in this study includes cases with a confirmed diagnosis by imaging, rather than only bleeding symptoms.
In this study, GI bleeding occurred in 2.5% of patients with critical COVID-19, which is not greater than that reported among general patients in intensive care units (4.7%).17 The results suggest that the incidence of GI bleeding from COVID-19 is not greater than that from other diseases. However, it is unclear from our results whether COVID-19 is more likely to cause GI bleeding than other diseases because this study did not compare bleeding rates between patients with and without COVID-19; therefore, it is not known whether the bleeding rate in patients with less severe disease is higher than that of other diseases.
Multivariate analysis showed an increased risk of GI bleeding with worsening severity of COVID-19 (table 3). Regarding the association between COVID-19 severity and GI symptoms, a report revealed that patients with severe COVID-19 experience more severe abdominal pain than those with non-severe COVID-1918; reports on the association of other GI symptoms and COVID-19 severity are limited. Meanwhile, only one study has reported the association between GI bleeding and COVID-19 severity.6 To our knowledge, this study is the first to reveal an association between COVID-19 severity and GI bleeding using multivariate analysis after considering potential confounders.
One factor influencing GI bleeding is the impact of posthospitalisation treatment. Among patients with severe COVID-19, there was no association between anticoagulant use (heparin, NOACs, and CHDF) and GI bleeding. In contrast, among patients with critical COVID-19, anticoagulant therapy (heparin, NOAC, CHDF, and ECMO) was significantly associated with GI bleeding. These results show that the higher OR in the critical group is probably because of strong anticoagulants such as those used in ECMO. Meanwhile, anticoagulation therapy was not associated with GI bleeding in patients with severe COVID-19, suggesting that in this category of patients, organ damage caused by COVID-19 itself is the risk for GI bleeding and not the anticoagulants used.
A possible mechanism of COVID-19-induced organ damage is through the ACE2, which is the target receptor of SARS-CoV-2 and is found in the adenocytes of the stomach, small intestine, colon, and rectal epithelium. Ischaemic changes associated with ACE2 receptor-mediated vascular endothelial damage may be involved in the development of GI symptoms.19 20 In the present study, endoscopic or CT studies performed to assess bleeding revealed cases of GI bleeding caused by ischaemic changes (eg, ischaemic enterocolitis) in the GI tract. GI disorders due to ischaemia occur more frequently in the lower GI tract, which has a poorer blood supply than the upper GI tract wherein ischaemic changes less frequently occur.21 Patients with COVID-19 have been reported to develop ischaemic GI disorders previously.22–27 In the cohort with GI bleeding due to pneumonia other than COVID-19, the rate of GI bleeding due to ischaemic changes was not high. It can be suggested that ischaemic changes associated with COVID-19 contributed to GI bleeding, resulting in an increased risk of lower GI bleeding but not upper GI bleeding.
In this study, PPI/P-cab use did not significantly reduce the risk of GI bleeding (table 3), which was still the case when the analysis was limited to upper GI bleeding (table 4). Additionally, the association between PPI/P-cab use and GI bleeding after hospitalisation showed that the rate of PPI/P-cab use was higher among patients with GI bleeding in the critical COVID-19 group (online supplemental file 12), suggesting that PPI/P-cab after hospitalisation may not be effective in preventing GI bleeding. A recent meta-analysis of patients in the intensive care unit demonstrated the usefulness of PPIs in preventing upper GI bleeding.28 Furthermore, the main causes of upper GI bleeding in this study were gastric ulcers and AGML (online supplemental file 5), which can be treated with PPIs. Hence, it is unclear why PPI use did not prevent upper GI bleeding in this study. However, a previous report on the association between COVID-19 and GI bleeding showed similar results.7 PPI-induced small intestinal damage has been reported29 and may be a cause of PPI ineffectiveness; however, only one case of small bowel inflammation was observed in the present study. Thus, it is unclear whether small bowel lesions influenced GI bleeding in this study. Future studies are needed to determine how GI bleeding can be prevented in patients with COVID-19.
This study has several limitations. First, this was a retrospective cohort study with participants limited to hospitals around Tokyo. Additionally, the small number of patients who experienced GI bleeding may cause bias and wide confidence intervals. However, since the bleeding rate in this study was comparable to that in a previous report,6 the bias is considered as not severe; however, further prospective studies are desirable. Second, although this study evaluated the use of anticoagulants, it was not able to determine whether anticoagulants were administered in therapeutic or prophylactic doses. Anticoagulant dosages may influence GI bleeding. Regarding anticoagulation therapy after hospitalisation, prophylactic doses were administered in most cases except for those who received strong anticoagulants, such as in ECMO or CHDF cases. Third, the severity of pneumonia was classified in this study according to the NIH classification, but there is a wide range of disease states even within the same severity level, such as the extent of pneumonia, which may affect the results. More detailed studies on each severity level are desirable in the future.
In conclusion, the severity of COVID-19 is a significant risk factor for GI bleeding in hospitalised patients. Critical COVID-19 is a risk factor for both upper and lower GI bleeding. In contrast, severe COVID-19 is a risk factor for lower GI bleeding but not for upper GI bleeding. Clinicians should be aware that patients with severe COVID-19 are at higher risk of GI bleeding.
Data availability statement
Data are available upon reasonable request.
Patient consent for publication
This study involves human participants and was approved by The Ethics Committee of the Tokyo Medical and Dental University (M2021-364) and research participation hospitals. Participants gave informed consent to participate in the study before taking part.
We thank the following people for their cooperation in collecting information from patient records: Kimie Suzuki (Tokyo Medical and Dental University Hospital), Kanae Metoki, Yuria Takei (Tokyo Metropolitan Hiroo Hospital), Maiko Motobayashi, Tomoaki Shirasaki, Naoko Oi, Yuki Matsui, Junko Yagita (Tokyo Metropolitan Otsuka Hospital), Chiaki Maeyashiki, Shun Ishido, Hiroaki Matsumoto, Tsubasa Nobusawa, Sakura Kirino, Yuka Hayakawa, Kento Inada, Ken Ueda, Hideki Kajiwara, Reiko Taki, Satoko Hanada (Musashino Red Cross Hospital), Yuko Karakama, Mamoru Watanabe (Tokyo Kyosai Hospital), Shunsuke Ueyama (Tsuchiura Kyodo General Hospital), Nahoko Mochizuki, Mariko Numata, Junko Fujiki, Shoko Toda, Sayaka Murata (Kashiwa Municipal Hospital), Masahiro Suzuki, Erika Aikawa, Mizuki Nagai, Chiaki Kamitsuma, Takashi Isozaki, Sho Kawasaki, Haruka Sato (JA Toride Medical Center), Shiori Ito (Ome Municipal General Hospital), Tomohiro Yamazaki and Shin Namiki (Tokyo Metropolitan Tama Medical Center). The authors would like to thank Enago (www.enago.jp) for the English language review.
This web only file has been produced by the BMJ Publishing Group from an electronic file supplied by the author(s) and has not been edited for content.
SH and TF contributed equally.
Contributors SH, TaF and ToF conceived of and designed the experiments. SH, TaF, SSuzuki, MK, SO, YF, SA, KT, HK, ST, MK, KN, FK, YI, TK, HI, SSakita, TY and HW reviewed the medical records and collected data. SH, AK and YM analysed the data. KO and RO supervised the study. SH, TaF and AK wrote the original manuscript draft. All authors contributed to the writing of the final manuscript. All authors approved the final version of the manuscript. ToF is the guarantor of this manuscript.
Funding The authors have not declared a specific grant for this research from any funding agency in the public, commercial or not-for-profit sectors.
Competing interests None declared.
Provenance and peer review Not commissioned; externally peer reviewed.
Supplemental material This content has been supplied by the author(s). It has not been vetted by BMJ Publishing Group Limited (BMJ) and may not have been peer-reviewed. Any opinions or recommendations discussed are solely those of the author(s) and are not endorsed by BMJ. BMJ disclaims all liability and responsibility arising from any reliance placed on the content. Where the content includes any translated material, BMJ does not warrant the accuracy and reliability of the translations (including but not limited to local regulations, clinical guidelines, terminology, drug names and drug dosages), and is not responsible for any error and/or omissions arising from translation and adaptation or otherwise.