Red Cell Distribution Width/Albumin Ratio: A Predictor of In-Hospital All-Cause Mortality in Patients with Acute Myocardial Infarction in the ICU

Introduction

Acute myocardial infarction (AMI), an inflammatory disease,1 remains the leading cause of cardiac death globally.2 Inflammation plays a pivotal role in the acute phase as well as the healing process following AMI.3 In the course of clinical practice, the vast majority of patients with AMI will be admitted to the intensive care unit (ICU)/ critical care unit (CCU). Routine ICU/CCU use for patients with AMI is recommended in Japanese and European guidelines.4,5 Although reduced by successful early reperfusion therapy, mortality is still substantial, particularly for patients who do not receive reperfusion therapy. Therefore, it is feasible to identify a convenient and effective inflammatory marker to identify high-risk patients with AMI admitted to the ICU and guide anti-inflammatory therapy.

Studies have shown that inflammatory markers are associated with the prognosis of AMI,6,7 and we speculated that complex markers may have advantages over single markers. RDW is a hematological parameter that reflects the size heterogeneity of red blood cells (RBCs) and is utilized for the differential diagnosis of anemia. As an easily available marker of systemic inflammation, a previous study also suggested that RDW is associated with poor prognosis in AMI.8–10 Serum albumin, a major protein in the human blood circulation system, is a conventional indicator of the nutritional and inflammatory status. Serum albumin content has been shown to be a remarkable predictor of adverse outcomes in individuals with AMI.11,12 The RDW/albumin ratio (RAR) is a newly identified novel risk marker, which is defined as the ratio of RDW to serum albumin value (%/g/dl). Studies have shown that bioindicators are important for determining the prognosis of diverse inflammatory diseases.13–15 RAR is a relatively new index measure; however, the established prognostic scoring systems for AMI do not include inflammatory indicators. Moreover, its application in people with AMI has not yet been investigated.

Thus, in this study, our objective was to evaluate the association between RAR and in-hospital mortality in patients with AMI admitted to the ICU, using real-world data.

Materials and Methods Data Source

The eICU Collaborative Research Database (CRD) is a large, open-access, multicenter critical care data resource created by the Massachusetts Institute of Technology (MIT) Laboratory for Computational Physiology in collaboration with Philips Healthcare between 2014 and 2015. The CRD contains data on over 200,000 admissions from over 200 hospitals across the United States. The Health Insurance Portability and Accountability Act safe harbor provisions allow data resources to be distributed without revealing the identity of any patient. As data in the eICU-CRD are de-identified, the requirement of informed consent is waived. This database was approved by the Institutional Review Board (IRB) of the MIT. This study was reviewed and approved by the Medical Ethics Committee of the First People’s Hospital of Changde (No. 2022–162-01). Furthermore, this study followed the Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) reporting guidelines, and was conducted in accordance with the ethical principles of the Declaration of Helsinki and its later amendments.

Study Population

We included adult patients with AMI based on the ICD-9 code 410, adopted in eICU-CRD. The following inclusion criteria were applied: i) age ≥18 years, ii) first ICU admission with a primary diagnosis of AMI, and iii) ICU length of stay >24 h. The exclusion criteria were as follows: i) age ≥89 years; ii) if key information (in-hospital mortality, body mass index [BMI], acute physiology and chronic health evaluation [APACHE] scores, hospital length of stay, ICU length of stay, and mechanical ventilation use) could not be obtained; iii) missing RDW and albumin measurements after ICU admission; and iv) blood transfusion received within 24 h before admission. The study population comprised 2594 patients with AMI [median age (interquartile range, IQR), 66.0, 57.0–76.0 years; female, 1634 (63.0%)]. The selection procedure for the study participants is summarized in Figure 1.

Figure 1 Flowchart of study participant selection. *Key information refers to mortality, body mass index (BMI), acute physiology and chronic health evaluation (APACHE) score, hospital length of stay, intensive care unit (ICU) length of stay, and mechanical ventilation use.

Data Extraction

One of the authors was granted permission to use the eICU-CRD data for medical research (certificate number 42039823). In this research, PostgreSQL v.11.2 (The PostgreSQL Global Development Group, https://www.postgresql.org/) was employed to extract patient data from the CRD. Clinical and laboratory findings of each patient were recorded on admission. The data, including demographic features, past medical history, type of AMI, APACHE scores, mortality, laboratory assessments, cause of admission, length of admission, and ICU stay, were obtained from the eICU-CRD following the first laboratory indicators after admission.

Definition of AMI

The term AMI is used when acute myocardial injury occurs, with clinical evidence of acute myocardial ischemia, and detection of a rise and/or fall of cTn values, with at least one value above the 99th percentile URL and at least one of the following:16

Symptoms of myocardial ischemia New ischemic electrocardiographic (ECG) changes Development of pathological Q waves Imaging evidence of new loss of viable myocardium or new regional wall motion abnormality in a pattern consistent with ischemic etiology Identification of coronary thrombi by angiography or at autopsyRAR Assessment and Outcomes

RDW is expressed as a percentage (%), and albumin in g/dL. The RAR was calculated as a ratio of RDW to albumin. In our study, the primary outcome was in-hospital mortality, while the secondary outcomes were mechanical ventilation use, length of ICU stay, and length of hospital stay. The observation period was from each patient’s first admission to the hospital until death.

Statistical Analysis

Continuous data are presented as median and IQR, whereas categorical variables are provided in terms of frequency and percentage. The outcome variable was used to stratify the baseline characteristics of all patients. To compare the groups, we employed the chi-square test for categorical variables and the Kruskal–Wallis test for continuous variables. Receiver operating characteristic (ROC) curve assessment was performed to test the capacity of RAR to differentiate between survivors and non-survivors and to determine the optimal RAR cutoff value. Additionally, RAR was compared with other prognostic indicators to verify its superiority. We performed a multivariate logistic regression analysis to determine whether RAR was related to in-hospital mortality in individuals with AMI, and the data were reported as odds ratios (OR) with 95% confidence interval (CI). In Model I, factors were adjusted for age, sex, and ethnicity. Age, sex, ethnicity, past medical history, AMI category, APACHE scores, white blood cell (WBC), RBC, hemoglobin (Hb), aspartate aminotransferase (AST), alanine aminotransferase (ALT), blood urea nitrogen (BUN), creatinine, total bilirubin (TB), total protein (TP), glucose, potassium, sodium, calcium, magnesium, bicarbonate, activated partial thromboplastin time (APTT), international normalized ratio (INR), troponin I, and natural log of B-type natriuretic peptide (lnBNP) were all adjusted for in Model II. Kaplan–Meier survival curves and the Log rank test were used to describe survival distributions. Stratified analysis was used to determine whether the impact of RAR was different in different subgroups. R (v3.42) and Empower Stats v2.17.8 (http://www.empowerstats.com/cn/) were used to analyze all data. All reported p values were two-sided, and two-tailed probability values of less than 5% were deemed statistically significant.

Results Characteristics of Patients

A total of 2594 patients with AMI were enrolled; of these, 2276 survived and 318 died during hospitalization. The baseline clinical features of the patients who died and those who survived are listed in Table 1. A remarkable number of patients who died had abnormal laboratory values, including greater WBC counts; higher blood glucose, RDW, creatinine, and ALT concentrations; and decreased RBC counts, total cholesterol, Hb, high-density lipoprotein (HDL), and albumin concentrations (p<0.01), as shown in Table 2. The average overall RAR was 4.18 (%g/dL). However, non-survivor patients had significantly higher RAR than the survivor group (5.04%/g/dL vs 4.09%/g/dL, p <0.001, Table 2).

Table 1 Baseline Clinical Features of the Enrolled Participants, Survivors, and Non-Survivors

Table 2 Baseline Laboratory Features in Survivors and Non-Survivors

Univariate analysis showed that RAR, age, chronic heart failure, APACHE score, mechanical ventilation use, WBC, RDW, creatinine, BUN, total bilirubin, potassium, calcium, bicarbonate, APTT, troponin I, and lnBNP were positively correlated with mortality. Female sex, hypertension, ST-segment elevation myocardial infarction (STEMI), RBC, Hb, albumin, TP, sodium, magnesium, glucose, and triglyceride levels were negatively correlated with mortality (Supplemexntary Table 1).

Relationship of RAR with Outcomes

We used ROC curve analysis to evaluate the RAR prediction ability and determine the optimal threshold to forecast in-hospital all-cause mortality in patients with AMI. Different ROC curves for RAR, RDW, albumin, APACHE score, and RAR combined with APACHE score were constructed. The AUCs (95% CI) for RAR, RDW, and albumin were 0.738 (0.720, 0.755); 0.624 (0.605, 0.643); and 0.696 (0.678, 0.714), respectively (Figure 2). The cutoff point for RAR was 4.776. We observed that RAR predicted in-hospital all-cause mortality better than RDW and albumin levels (p<0.01). The AUC (95% CI) of RAR combined with APACHE score and APACHE score was 0.860 (0.845, 0.872) and 0.841 (0.833, 0.861), respectively (p<0.01) (Figure 3). These results indicate that RAR combined with the APACHE score has a better predictive power than the APACHE score alone.

Figure 2 Receiver-operating characteristic curve of RAR, albumin, and RDW to predict in-hospital all-cause mortality of AMI.

Abbreviations: RAR, red cell distribution width/albumin ratio; RDW, red cell distribution width; AMI, acute myocardial infarction.

Figure 3 Receiver-operating characteristic curve of RAR and RAR combined with APACHE score to predict in-hospital all-cause mortality.

Abbreviations: RAR, red cell distribution width/albumin ratio; APACHE, Acute Physiology and Chronic Health Evaluation.

The patients were divided into two sets according to the cutoff point. Greater RARs (≥4.776%/g/dL) were related to longer hospital and ICU stay, higher rates of mechanical ventilation use, and a higher in-hospital mortality (P <0.001, Table 3). We further explored these findings using multivariate logistic regression analysis, as shown in Table 4. After adjusting for the clinical confounders listed, RAR levels (per 1%/g/dL increase) were related to a 27% higher risk of in-hospital mortality (OR 1.27, 95% CI, 1.12, 1.43, p <0.001). Compared to the lower RAR group (<4.776%/g/dL), patients in the higher RAR group (≥4.776%/g/dL) showed elevated in-hospital mortality (OR 1.62, 95% CI 1.05, 2.51, p=0.03) in the adjusted Model II. A similar relationship was observed for the secondary outcomes.

Table 3 Outcomes of the Patients with AMI Across Cutoff Points of the RAR

Table 4 Results of Multivariable Logistic Regression of RAR and In-Hospital All-Cause Mortality

Subgroup assessment was performed to explore the relationship between RAR and in-hospital mortality rate (Table 5). No remarkable interaction was observed across the strata (p =0.0800–0.7854).

Table 5 Subgroup Assessments of the Associations of In-Hospital All-Cause Mortality with RAR

Figure 4 shows the Kaplan–Meier survival curves for the RAR. We found that patients with high levels of RAR (≥4.776%/g/dL) had remarkably poorer survival than those with low levels (<4.776%/g/dL) (p <0.0001).

Figure 4 Kaplan–Meier curve of RAR for estimating in-hospital all-cause mortality of AMI.

Abbreviations: RAR, red cell distribution width/albumin ratio; AMI, acute myocardial infarction.

Discussion

This study illustrated that the RAR is an independent predictor of the mortality of AMI patients in the ICU and is positively associated with in-hospital all-cause mortality. We confirmed that the composite indicator has a higher predictive power for in-hospital mortality compared to albumin or RDW.

RDW is a parameter exhibiting the heterogeneity of RBC volume, which represents the variability in the size of circulating RBCs, and is an easily obtainable and inexpensive blood marker.17,18 Previous investigations have shown that RDW is linked to acute coronary syndrome (ACS);9,10,17,19,20 however, the mechanisms underlying this association are not completely understood. RDW is considered an inflammatory marker that can worsen arteriosclerosis and is associated with several inflammatory markers.20,21 In atherosclerosis, RBCs are influenced by inflammatory factors, which trigger the generation of numerous immature RBCs from the bone marrow, resulting in higher RDW levels and elevated ineffective hematopoiesis in the bone marrow.22–24 The rupture of the unstable fibrous cap causes thrombosis, and the deformability of RBCs trapped within a fibrin clot decreases, resulting in an increased RDW.18 The study by Uyarel et al19 demonstrated that higher RDW values at baseline were more remarkably linked to a higher cardiovascular mortality among patients with STEMI undergoing primary percutaneous coronary intervention (pPCI) both in-hospital and (1.8±1.3 years) after AMI. In a median follow-up time of 27 months, a recent study illustrated that an RDW increase during hospitalization was likewise related to eventual death (adjusted hazard ratio [HR] 1.13 for 1-standard deviation [SD] increase in RDW, 95% CI 1.02–1.25) in 1709 patients with AMI.10 In individuals with and without anemia, there was a link between a greater RDW and worse outcomes. Similarly, in our study, the non-survivor group had higher RDW levels than the survivor group.

Serum albumin levels were routinely measured in patients with AMI, and albumin also plays a role in the acute early inflammatory response.25–28 Nutritional intake and systemic inflammation influence albumin production rates.29 Inflammatory cytokines such as interleukin (IL)-6 and tumor necrosis alpha (TNF-α) restrict albumin production by inhibiting albumin gene transcription, resulting in decreased albumin synthesis. Previous investigations11,30,31 revealed that hypoalbuminemia is linked to adverse events in individuals with ACS. Moreover, one of these studies indicated that albumin content ≤3.50 g/dL is an independent prognostic factor for new-onset heart failure and in-hospital mortality in individuals with ACS. Plakht et al32 reported that decreased albumin level on admission, consisting of levels within “normal” clinical range, is remarkably linked to long-term all-cause mortality in hospital survivors of AMI with a “dose-response” type association. Comparable results were obtained in our study, showing that the albumin contents were lower in patients who died.

Several studies exist on composite indicators for AMI. In most studies, composite indicators are better than single-characteristic indicators. According to Tong et al,33 RDW-to-platelet ratio is an independent predictor of in-hospital mortality in AMI patients. In our study, the RAR predicted in-hospital mortality in AMI patients in the ICU better than the RDW or albumin alone. RAR combined with the APACHE score had better predictive power than the APACHE score alone. This study also found that the optimal cutoff value of RAR for predicting in-hospital mortality was 4.776, which is similar to that in another report (4.59 in patients with acute respiratory distress syndrome14). Our study demonstrates, for the first time, that RAR ≥4.776 is associated with increased in-hospital mortality in patients with AMI in the ICU. We also found that RAR is associated with a higher proportion of patients requiring mechanical ventilation use and longer ICU stay. Therefore, it is essential to carry out routine blood and albumin tests at admission. Recently, RAR was found to be a strong predictor of mortality in individuals with acute respiratory distress syndrome and diabetic ketoacidosis.14,15 Increased RDW and low serum albumin levels are closely associated with the inflammatory response. RAR can be performed easily and rapidly in laboratories. As a biomarker, it reflects inflammatory processes in AMI and can be used to evaluate the short-term outcome of AMI. In conclusion, our data illustrate that RAR is a predictor of in-hospital mortality among ICU patients with AMI.

Limitations

This study has some limitations. First, due to the retrospective status of the work, selection bias cannot be totally excluded. Second, this study was limited by the fact that RAR levels were assessed only once and were not evaluated for their dynamics changes. Third, other unstudied confounders may have interfered with our results. Owing to the limitations of the eICU database, we could not obtain comprehensive hospitalization data of the patients, such as cardiac function classification, echocardiogram results, specific location of myocardial infarction, thrombolysis in myocardial infarction (TIMI) risk score, Global Registry of Acute Coronary Events (GRCAE) risk score, and other data that may affect the prognosis of patients. Further studies are required to verify the results of this study. Despite these shortcomings, our study is meaningful in exploring the relationship between RAR and AMI.

Conclusions

We concluded that the RAR is a potential biomarker of AMI, and elevated RAR is linked to increased in-hospital mortality in individuals with AMI in the ICU. RAR is a more accurate predictor of in-hospital all-cause mortality in patients with AMI in the ICU than albumin or RDW.

Data Sharing Statement

The eICU data resource (version 2.0) is freely accessible at https://physionet.org/content/eicu-crd/. Access to the data resource is granted to any researcher who follows the data usage guidelines.

Ethics Approval and Consent to Participate

This database was approved by the Institutional Review Board of the Massachusetts Institute of Technology. The Medical Ethics Committee of The First People’s Hospital of Changde has approved the research (No. 2022-162-01). It was a retrospective analysis of data from a third-party anonymized and publicly accessible data resource; informed consent from the patients was not required.

Acknowledgments

The authors express their gratitude to the researchers at the MIT Laboratory for Computational Physiology and other collaborators who maintain the eICU-CRD datasets up to date.

Disclosure

The authors declared that they have no conflicts of interest to this work.

References

1. Damar İH, Eroz R. The association of hereditary prothrombotic risk factors with ST-elevation myocardial infarction. Medeni Med J. 2020;35:295–303. doi:10.5222/MMJ.2020.67366

2. Santos-Gallego CG, Picatoste B, Badimón JJ. Pathophysiology of acute coronary syndrome. Curr Atheroscler Rep. 2014;16(4):401. doi:10.1007/s11883-014-0401-9

3. Tomandlova M, Jarkovsky J, Tomandl J, et al. Prognostic value of pentraxin-3 level in patients with STEMI and its relationship with heart failure and markers of oxidative stress. Dis Markers. 2015;2015:159051. doi:10.1155/2015/159051

4. Ibanez B, James S, Agewall S, et al. 2017 ESC guidelines for the management of acute myocardial infarction in patients presenting with ST-segment elevation: the task force for the management of acute myocardial infarction in patients presenting with ST-segment elevation of the European Society of Cardiology (ESC). Eur Heart J. 2018;39:119–177. doi:10.1093/eurheartj/ehx393

5. Kimura K, Kimura T, Ishihara M, et al. JCS 2018 guideline on diagnosis and treatment of acute coronary syndrome. Circ J. 2019;83(5):1085–1196. doi:10.1253/circj.CJ-19-0133

6. Ries W, Torzewski J, Heigl F, et al. C-reactive protein apheresis as anti-inflammatory therapy in acute myocardial infarction: results of the CAMI-1 study. Front Cardiovasc Med. 2021;8:591714. doi:10.3389/fcvm.2021.591714

7. Qi LY, Liu HX, Cheng LC, et al. Prognostic value of the leuko-glycemic index in acute myocardial infarction patients with or without diabetes. Diabetes Metab Syndr Obes. 2022;9(15):1725–1736. doi:10.2147/DMSO.S356461

8. Cemin R, Donazzan L, Lippi G, et al. Blood cells characteristics as determinants of acute myocardial infarction. Clin Chem Lab Med. 2011;49(7):1231–1236. doi:10.1515/CCLM.2011.183

9. Ilhan E, Güvenç TS, Altay S, et al. Predictive value of red cell distribution width in intrahospital mortality and postintervention thrombolysis in myocardial infarction flow in patients with acute anterior myocardial infarction. Coron Artery Dis. 2012;23(7):450–454. doi:10.1097/MCA.0b013e3283587897

10. Dabbah S, Hammerman H, Markiewicz W, et al. Relation between red cell distribution width and clinical outcomes after acute myocardial infarction. Am J Cardiol. 2010;105(3):312–317. doi:10.1016/j.amjcard.2009.09.027

11. González-Pacheco H, Luis M, Sandoval J, et al. Prognostic implications of serum albumin levels in patients with acute coronary syndromes. Am J Cardiol. 2017;119(7):951–958. doi:10.1016/j.amjcard.2016.11.054

12. Zhu L, Chen M, Lin X. Serum albumin level for prediction of all-cause mortality in acute coronary syndrome patients: a meta-analysis. Biosci Rep. 2020;40(1):BSR20190881. doi:10.1042/BSR20190881

13. Chen L, Lu XY, Zhu CQ. Prognostic value of albumin-red cell distribution width score in patients with severe community-acquired pneumonia. Ann Palliat Med. 2020;9(3):759–765. doi:10.21037/apm.2020.04.22

14. Yoo J-W, Sunmi J, Lee SJ, et al. Red cell distribution width/albumin ratio is associated with 60-day mortality in patients with acute respiratory distress syndrome. Infect Dis. 2020;52(4):266–270. doi:10.1080/23744235.2020.1717599

15. Zhou D, Wang J, Xiaokun L. The red blood cell distribution width-albumin ratio was a potential prognostic biomarker for diabetic ketoacidosis. Int J Gen Med. 2021;14:5375–5380. doi:10.2147/IJGM.S327733

16. Thygesen K, Alpert JS, Jaffe AS, et al. Executive group on behalf of the Joint European Society of Cardiology (ESC)/American College of Cardiology (ACC)/American Heart Association (AHA)/World Heart Federation (WHF) Task Force for the Universal Definition of Myocardial Infarction. Fourth Universal Definition of Myocardial Infarction (2018). J Am Coll Cardiol. 2018;72(18):2231–2264. doi:10.1016/j.jacc.2018.08.1038

17. Arbel Y, Birati EY, Finkelstein A, et al. Red blood cell distribution width and 3-year outcome in patients undergoing cardiac catheterization. J Thromb Thrombolysis. 2014;37(4):469–474. doi:10.1007/s11239-013-0964-2

18. Salvagno GL, Sanchis-Gomar F, Picanza A, et al. Red blood cell distribution width: a simple parameter with multiple clinical applications. Crit Rev Clin Lab Sci. 2015;52(2):86–105. doi:10.3109/10408363.2014.992064

19. Uyarel H, Ergelen M, Cicek G, et al. Red cell distribution width as a novel prognostic marker in patients undergoing primary angioplasty for acute myocardial infarction. Coron Artery Dis. 2011;22(3):138–144. doi:10.1097/MCA.0b013e328342c77b

20. Azab B, Torbey E, Hatoum H, et al. Usefulness of red cell distribution width in predicting all-cause long-term mortality after non-ST-elevation myocardial infarction. Cardiology. 2011;119(2):72–80. doi:10.1159/000329920

21. Lippi G, Targher G, Montagnana M, et al. Relation between red blood cell distribution width and inflammatory biomarkers in a large cohort of unselected outpatients. Arch Pathol Lab Med. 2009;133(4):628–632. doi:10.5858/133.4.628

22. Hansson GK. Inflammation, atherosclerosis, and coronary artery disease. N Engl J Med. 2005;352(16):1685–1695. doi:10.1056/NEJMra043430

23. Yčas JW, Horrow JC, Horne BD. Persistent increase in red cell size distribution width after acute diseases: a biomarker of hypoxemia? Clin Chim Acta. 2015;448:107–117. doi:10.1016/j.cca.2015.05.021

24. Allen LA, Michael Felker G, Mehra MR, et al. Validation and potential mechanisms of red cell distribution width as a prognostic marker in heart failure. J Card Fail. 2010;16(3):230–238. doi:10.1016/j.cardfail.2009.11.003

25. Gabay C, Kushner I. Acute-phase proteins and other systemic responses to inflammation. N Engl J Med. 1999;340(6):448–454. doi:10.1056/NEJM199902113400607

26. Eckart A, Struja T, Kutz A, et al. Relationship of nutritional status, inflammation, and serum albumin levels during acute illness: a prospective study. Am J Med. 2020;133(6):713–722. doi:10.1016/j.amjmed.2019.10.031

27. Mukai H, Villafuerte H, Qureshi AR, et al. Serum albumin, inflammation, and nutrition in end-stage renal disease: c-reactive protein is needed for optimal assessment. Semin Dial. 2018;31(5):435–439. doi:10.1111/sdi.12731

28. Bozkurt E, Ucak T. Serum inflammation biomarkers in patients with keratoconus. Ocul Immunol Inflamm. 2020;7:1–4.

29. Suzuki S, Hashizume N, Kanzaki Y, et al. Prognostic significance of serum albumin in patients with stable coronary artery disease treated by percutaneous coronary intervention. PLoS One. 2019;14(7):e0219044. doi:10.1371/journal.pone.0219044

30. Hartopo AB. Putrika Prastuti Ratna Gharini, Budi Yuli Setianto. Int Heart J. 2010;51(4):221–226.

31. Polat N, Oylumlu M, Işik MA, et al. Prognostic Significance of Serum Albumin in Patients With Acute Coronary Syndrome. Angiology. 2020;71(10):903–908. doi:10.1177/0003319720941747

32. Plakht Y, Gilutz H, Shiyovich A. Decreased admission serum albumin level is an independent predictor of long-term mortality in hospital survivors of acute myocardial infarction. Int J Cardiol. 2016;219:20–24.

33. Tong L, Liu Y, Shen JH, et al. Relationship between the red cell distribution width-to-platelet ratio and in-hospital mortality among critically ill patients with acute myocardial infarction: a retrospective analysis of the MIMIC-IV database. BMJ Open. 2022;12:e062384. doi:10.1136/bmjopen-2022-062384

留言 (0)

沒有登入
gif