Open access Protocol Evaluation of the role of sex as a prognostic factor in critically ill adults with sepsis: systematic review protocol Jesus Lopez-­Alcalde ‍ ‍,1,2,3 Alba Antequera Martín ‍ ‍,4,5 Elena Stallings,2 Alfonso Muriel ‍ ‍,2,3,6 Borja Fernández-­Félix,2,3 Ivan Solà ‍ ‍,5,7,8 Rosa del Campo ‍ ‍,9 Manuel Ponce-­Alonso ‍ ‍,9 Federico Gordo,1,10 Pilar Fidalgo,1,11 Ana Verónica Halperin,9 Noelia Álvarez-­Díaz ‍ ‍,12 Olaya Madrid-­Pascual ‍ ‍,13 Gerard Urrutia,5,7,8 Javier Zamora ‍ ‍2,3 To cite: Lopez-­Alcalde J, Abstract Antequera Martín A, Stallings E, Strengths and limitations of this study Introduction Sepsis is a leading cause of mortality in et al. Evaluation of the critically ill patients. Recently, it has been recognised that role of sex as a prognostic ►► To our knowledge, this systematic review will be the sex may contribute to a differential risk for developing factor in critically ill adults first addressing sex as a prognostic factor for sep- sepsis and it remains uncertain if the prognosis of sepsis with sepsis: systematic sis and its findings may help to stratify patients for review protocol. BMJ Open varies between the sexes. The aim of this systematic treatment. 2020;10:e035927. doi:10.1136/ review is to summarise the available evidence to assess ►► This systematic review will examine mortality out- bmjopen-2019-035927 the role of sex as a prognostic factor in patients with comes, which are critical for patients and decision sepsis managed in the intensive care unit (ICU). ►► Prepublication history and making. Methods and analysis This is a systematic review additional material for this ►► There may be challenges for meta-­analysing the re- protocol of prognostic studies of sex in patients with paper are available online. To sults due to heterogeneity among primary studies. view these files, please visit sepsis managed in the ICU. The primary outcomes include the journal online (http://​dx.​doi.​ all-­cause hospital mortality and all-­cause hospital mortality org/​10.​1136/​bmjopen-​2019-​ during the first 28 days. The secondary outcomes include 035927). all-­cause hospital mortality during the first 7 days and making it the main cause of admission to an all-­cause mortality at 1 year. We will conduct a search intensive care unit (ICU) in high-­ income JL-­A and AAM contributed strategy based on the population (sepsis), the prognostic countries.2 3 Moreover, despite advances in equally. factor (sex), the outcome of interest (mortality) and healthcare, sepsis remains a leading cause prognostic study methods. We will search in the following of death every year and is the most common Received 22 November 2019 Revised 21 February 2020 databases up to December 2019: MEDLINE Ovid (from cause of death within the ICU.4–6 Accepted 07 April 2020 1976), Embase Elsevier (from 1974), Web of Science Sepsis is a heterogeneous disease affecting and two trial registries. We will impose no language both male and female individuals.7 Recently, restrictions. Two authors will independently screen titles, it has been recognised that sex may contribute abstracts and full-­text articles for eligibility of studies, and subsequently extract data. Two authors will independently to a differential risk for developing sepsis; and assess the risk of bias of each study using the Quality it remains uncertain if the prognosis of sepsis in Prognostic Studies (QUIPS) tool. If possible, we will varies between the sexes.8–14 Prognosis factors carry out a meta-­analysis to provide a pooled prognostic are relevant in clinical care as they can iden- © Author(s) (or their effect estimate for each outcome. We will use the Grading tify risk groups in which clinical practice may employer(s)) 2020. Re-­use of Recommendations Assessment, Development and be tailored to the aim of reducing morbidity permitted under CC BY-­NC. No commercial re-­use. See rights Evaluation system to assess the quality of evidence. and mortality.15 The aim of this systematic and permissions. Published by Ethics and dissemination Ethical approval will not be review is to summarise the available evidence BMJ. required. Findings from this review will be reported in a to assess the role of sex as a prognostic factor For numbered affiliations see peer-­reviewed scientific journal. Additionally, the results for mortality in patients with sepsis managed end of article. will be disseminated at conferences and in the mass in the ICU. media. Correspondence to PROSPERO registration number CRD42019145054. Dr Alba Antequera Martín, Department of Epidemiology Methods and Public Health, Biomedical Introduction Registration Research Institute Sant Pau. Sepsis is currently defined as life-­threatening We registered the protocol with an Inter- Sant Antoni Maria Claret 167, organ dysfunction resulting from a dysregu- national Prospective Register of Systematic building 18, 08025 Barcelona, Spain; lated host response to inflammation.1 Sepsis Reviews (PROSPERO) on 12 November ​alba.​antequera.​martin@​gmail.​ is a major health problem with a worldwide 2019. Possible amendments will be updated com incidence of about 1 case per 1000 patients, on PROSPERO. The protocol is reported Lopez-­Alcalde J, et al. BMJ Open 2020;10:e035927. doi:10.1136/bmjopen-2019-035927 1 Open access according to the recommendations of the Preferred Information sources Reporting Items for Systematic Review and Meta-­Analysis We will search the following databases: (PRISMA) Protocols statement.16 The review will be ►► MEDLINE Ovid (from 1946 onwards). reported according to the PRISMA statement.17 ►► Embase Elsevier (from 1974 onwards). ►► Web of Science (WoS). Eligibility criteria Search methods Type of studies The search strings will be based on terms related to the We will include experimental studies, randomised or not, population (sepsis), the prognostic factor (sex), the and any observational study investigating the prognostic outcome (mortality) and prognostic study methods.27 significance of sex in adults with sepsis managed in the For ‘sex’ we will use a search string adapted from Loren- ICU. Regarding prognostic factor studies, we will only zetti, Moerman, Song and Stewart28–31(online supplemen- include phase II confirmatory studies. tary appendix 2). We will not apply publication date or language restrictions. Type of participants Moreover, we will check the bibliographic references Adults (male and female individuals, which allow deter- for additional relevant studies. We will also scan ​Clinical- mining the role of sex) aged ≥16 years with a diagnosis Trials.​ gov (​www.​clinicaltrials.​ gov) and the WHO Inter- of sepsis. We will accept the study authors’ definition of national Clinical Trials Registry Platform (​apps.​who.​int/​ sepsis and septic shock. For a study to be eligible, the trialsearch/) for unpublished and ongoing studies. sepsis must be managed in an ICU. Studies including Furthermore, we will search for conference abstracts of adults, children and/or adolescents will be eligible if major critical care and infectious diseases symposia from adults represent >80% of the study sample. 2010. 1. Interscience Conference on Antimicrobial Agents and Type of prognostic factors Chemotherapy (ICAAC); 50th edition 2010 to 59th We will include studies that assess sex as a prognostic edition 2019. factor. We will accept any assessment of sex as provided 2. European Congress of Clinical Microbiology and In- by the study authors. The concepts of sex and gender fectious Diseases (ECCMID); 20th edition 2010 to 29th are distinct. Sex is a biological characteristic and gender edition 2019. refers to the socially constructed roles, behaviours and 3. Society for Healthcare Epidemiology of America identities of women, men and gender-­ diverse individ- (SHEA): IDWeek 2012 to 2019 editions. uals.18 This review will not assess the association between 4. International Conference on Prevention and Infection gender and the outcomes of patients with sepsis in the Control (ICPIC): 2011, 2013, 2015, 2017, 2019 ICU. 5. Society of Critical Care Medicine (SCCM): 39th edi- We will assess the role of sex as a prognostic factor tion 2010 to 48th edition 2019. after adjustment for additional covariates. After a liter- 6. International Symposium on Intensive Care and Emer- ature search and a consensus process with clinician gency Medicine (ISICEM): 30th edition 2010 to 39th review authors, and as suggested during peer review of edition 2019. our protocol (online supplementary appendix 1), we 7. European Society of Intensive Care Medicine (ESICM): agreed on the following core set of adjustment factors: 2010 onwards. age, severity score (Sequential Organ Failure Assessment Apart from that, we will contact experts on the topic to socre (SOFA), Simplified Acute Physiology Score II (SAPS ask for any ongoing, missing or unreported studies. II) or Acute Physiologic Assessment and Chronic Health Evaluation II (APACHE II)), comorbidities (immunosup- pression,pulmonary diseases, cancer, liver diseases and Study records alcohol dependence), non-­urinary source of infection, Data management inappropriate or late antibiotic coverage.19–26 We may We will eliminate duplicates with Endnote 7.8 reference modify this list if we find new evidence that justifies the management software.32 We will manage the resultant changes. unique references with the online software EPPI-­Reviewer to identify additional duplicates, select studies, build the Type of outcomes data extraction templates and extract the data.33 Primary outcomes ►► All-­cause hospital mortality (the longest follow-­ up Study selection and data collection provided by the study authors). Two of six review authors (AAM, AH, MPA, OM, RdC, PF) ►► All-­cause hospital mortality during the first 28 days. will independently screen retrieved results according to the best practice guidelines to determine whether they Secondary outcomes fulfil the eligibility criteria of this review.34 We will use a ►► All-­cause hospital mortality during the first 7 days. consensus method, and a third author (JL-­A, AAM, ES) ►► All-­cause mortality at 1 year. will be consulted if there are disagreements. We will pilot 2 Lopez-­Alcalde J, et al. BMJ Open 2020;10:e035927. doi:10.1136/bmjopen-2019-035927 Open access eligibility criteria on a sample of references to ensure We will make a judgement for every criterion, using one their reproducibility. Assessment of the eligibility of full of four categories (low, moderate, high risk and unclear), texts will be undertaken similarly. Two of seven authors adapting the QUIPS guidelines.35 37 38 We will resolve (JL-­A, AAM, AH, MP, OM, RdC, PF) will independently disagreements by discussion, and a third author will be extract data according to a pre-­defined form and a third consulted if it is required. author will resolve the discrepancies (JL-­A, AAM, ES). We We will evaluate the impact of risk of bias from primary will pilot the data extraction template with three studies studies by performing sensitivity analyses (see the Sensi- to ensure its suitability. We will compare results and tivity analyses section). resolve discrepancies by discussion or by referring to a third review author. Data synthesis For each study and prognostic effect estimate, we will Data items extract the measures of association reported in each We will use CHecklist for critical Appraisal and data study (OR, RR, HR) along with its standarderror (SE) extraction for systematic Reviews of prediction Modelling or confidence intervals (CIs). We will transform associa- tion measures into OR with its 95% CI in order to allow Studies for prognostic factor (CHARMS-­PF) guidance for statistical pooling whenever adequate.39 We acknowledge data collection.35 This tool modifies the original CHARMS that the studies providing the adjusted prognostic effect checklist for prediction model studies.36 We will extract of a particular factor can differ in the set of adjustment the following data from each eligible study: factors. We will define a core set of adjustment factors for 1. Publication date, country and setting in which the each review outcome. study was conducted. If the study provides a set of adjustment factors that 2. Study design: experimental study or observational differ from our core set, we will meta-­analyse the study study. but the estimate will be penalised as part of the risk of bias 3. Registration: presence of registration or protocol assessment. If the same study presents different estimates published (yes/no). for the same outcome, each of them adjusted for different 4. Sepsis definition used by authors. factors, we will extract the estimate that has adjusted for 5. Participant characteristics: participant description, the maximum number of our key confounders for meta-­ eligibility criteria and details of treatment received (if analysis. If there are several estimations, all of them having applicable). adjusted for the same key confounders, we will consider 6. Sex definition used by authors. the estimate adjusted for more confounders in total. We 7. Review outcomes reported in the study: mortality; assume that this will minimise the risk of confounding type of measures (binary); follow-­up period (the first bias in the estimation. 7 days, the first 28 days and the longest follow-­up pro- We plan to combine the results from individual studies vided by the study authors). in a meta-­analysis to provide a pooled prognostic effect 8. Missing data: attrition (quantification and reasons), estimate for each outcome. We will carefully evaluate the handling of missing data by the study authors(avail- censoring mechanisms assumed in the studies that have able case analysis or imputation). been analysed using time-­to-­event procedures (i.e., Cox 9. Analysis: univariate analysis/logistic regression/Cox proportional hazard models). If we conclude that there regression/other. is a risk of bias due to informative censoring (e.g., when 10. Estimates reported between the prognostic factor patients with better prognosis are discharged for the and each review outcome: (a) unadjusted estimate: ICU), we will disregard HRs obtained in such studies. We association between sex and mortality with no covari- plan to use the Hartung-­Knapp-­Sidik-­Jonkman (HKSJ) ate; (b) adjusted estimates: association/s between sex method for random-­ effects meta-­analysis. We will use and mortality with at least one covariate. the Cochrane statistical package Review Manager 5.3 for 11. Type of measure of association: odds ratio (OR), risk organising the text of the review, and if the meta-­analysis ratio (RR) or hazard ratio (HR). is possible, we will use the metareg command in Stata 12. List ofadjustment factors that were used. We will con- (Stata V.15.1) for performing the meta-­analysis with the tact authors of individual studies for additional infor- HKSJ method.40 mation if required. Heterogeneity between studies Assessment of methodological quality and risk of bias When there are at least three studies, we will examine We will use the QUIPS (Quality In Prognosis Studies) heterogeneity by calculating prediction intervals in order tool to assess risk of bias.35 Two review authors will inde- to inform how the effect varies across populations. The pendently assess the following six domains of the QUIPS prediction interval is an index of dispersion that describes tool: (1) study participation; (2) study attrition; (3) prog- the range of effect sizes in which we would expect that nostic factor measurement; (4) outcome measurement; the true effect size will fall.41 We will use the spreadsheet (5) adjustment for other prognostic factors; (6) statistical provided by Borenstein et al for computing the prediction analysis and reporting. interval.42 Lopez-­Alcalde J, et al. BMJ Open 2020;10:e035927. doi:10.1136/bmjopen-2019-035927 3 Open access If we judge that the meta-­analysis is inappropriate for Author affiliations 1 other reasons, we will not combine results and we will Faculty of Health Sciences, Universidad Francisco de Vitoria, Pozuelo de Alarcón, Spain undertake a narrative analysis of studies. 2 Clinical Biostatistics Unit, Instituto Ramón y Cajal de Investigación Sanitaria, To better understand the sources of heterogeneity, we Madrid, Spain plan to conduct the following subgroup analyses: 3 Epidemiology and Public Health Networking Biomedical Research Centre ►► Trials versus cohort studies versus case–control studies. (CIBERESP), Madrid, Spain 4 ►► Prospective studies (all the study parts were prospec- Autonomous University of Barcelona, Barcelona, Spain 5 Biomedical Research Institute Sant Pau, Hospital Santa Cruz y San Pablo, tive) versus retrospective studies (all the study parts Barcelona, Spain were retrospective). 6 Department of Nursing and Physiotherapy, Universidad de Alcalá, Madrid, Spain 7 Epidemiology and Public Health Networking Biomedical Research Centre Sensitivity analyses (CIBERESP), Barcelona, Spain We will undertake the following sensitivity analyses: 8 Iberoamerican Cochrane Centre, Barcelona, Spain ►► Evaluate the impact of risk of bias from individual 9 Department of Microbiology, Hospital Universitario Ramón y Cajal, Madrid, Spain 10 trials in the overall effect. We will consider the Department of Intensive Care, Hospital Universitario del Henares, Coslada, Madrid, Spain following risk of bias domains as key domains for 11 Department of Internal Medicine, Hospital Universitario del Henares, Coslada, the analyses: study attrition, prognostic factor meas- Madrid, Spain urement, outcome measurement and adjustment for 12 Medical Library, Hospital Universitario Ramón y Cajal, Madrid, Madrid, Spain other prognostic factors. 13 Arztpraxis Kalkbreite, Zurich, Switzerland ►► We will exclude studies with a high risk of bias in at least one key domain. Twitter Jesus Lopez-­Alcalde @JLopezAlcalde ►► We will exclude studies with a high or moderate risk of Acknowledgements The authors thank Centro de Investigación Biomédica en Red bias in at least one key domain. (CIBER) for its support. ►► Assess the effect of excluding studies that have Contributors Guarantor: JL-­A. Conceiving of the review: JZ, JL-­A, AAM. Designing provided an adjusted estimate, but that did not adjust protocol: JL-­A, AAM, ES, JZ, IS. PROSPERO registration: AAM, JL-­A. Designing research strategies: NA-­D, IS, JL-­A, AAM. Drafting of the manuscript; contributed for all our core set of additional prognostic factors. equally to this work as first authors: AAM, JL-­A. Methodological approach: JL-­A, AAM, AM, IS, JZ, ES, GU. Clinical perspective: RdC, MP-­A, FG, PF, AH, AAM. Publication bias Screening search results: RdC, MP-­A, PF, AH, OM-­P, AAM, JL-­A, FG. Analysing We will attempt to assess the presence of publication bias data: AM, AAM, JL-­A, JZ, BF-­F. Interpreting data: AAM, JL-­A, AM, IS, JZ. Securing for each meta-­analysis containing ≥10 studies by funnel funding for the review: JL-­A, JZ, GU. Critical revision of the manuscript for important intellectual content: All authors. Alba Antequera Martín is a doctoral candidate plot representation and Peter’s test at a 10% level.43 in Public Health and Methodology of Biomedical Research, at the Department of Pediatrics, Obstetrics, Gynaecology and Preventive Medicine at Universitat Confidence in cumulative evidence Autònoma de Barcelona (Spain) and this work is part of her PhD. We will attempt to use the Grading of Recommendations Funding The SEXCOMPLEX project was supported by Instituto de Salud Carlos Assessment, Development and Evaluation (GRADE) III (Plan Estatal de I + D + i 2013–2016) and co-­financed by the European Development Regional Fund 'A way to achieve Europe' (ERDF) grant number approach and will present our results for each main PIE16/00050. AAM was funded by the Instituto de Salud Carlos III through the outcome in a ‘Summary of findings’ table, although “AcciónEstratégica en Salud 2013–2016 / Contratos Río Hortega call 2018/ we acknowledge that the GRADE approach has not CM18/00141” (Co-­funded by European Social Fund 2014–2020, “Investing in your been widely used to assess the certainty of the evidence future”). BF-­F was funded by Networking Biomedical Research Centre (CIBER). Publishing costs were provided by Networking Biomedical Research Centre (CIBER). of prognostic factors.44 45 We will follow the principles These funding sources had no role in the design of this review, its execution, suggested in Hughet et al and Iorio et al for prognostic analyses, interpretation of the data, or decision to submit results. questions.46 47 We will use GRADE profiler (GRADEpro Disclaimer The funding sources had no role in the design of this review, its GDT) for designing a ‘Summary of findings’.48 execution, analyses, interpretation of the data or decision to submit results. Competing interests None declared. Patients and public involvement Patient and public involvement Patients and/or the public were not involved in Patients and the general public will not be involved in the the design, or conduct, or reporting, or dissemination plans of this research. process of conducting this systematic review. We will not Patient consent for publication Not required. plan to involve patients in the dissemination of the results Provenance and peer review Not commissioned; externally peer reviewed. of this systematic review. Open access This is an open access article distributed in accordance with the Creative Commons Attribution Non Commercial (CC BY-­NC 4.0) license, which Ethics and dissemination permits others to distribute, remix, adapt, build upon this work non-­commercially, Ethical permission is not necessary as this study summarises 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 published data. This review is part of the SEXCOMPLEX is non-­commercial. See: http://​creativecommons.​org/​licenses/​by-​nc/​4.​0/. project (Influence of sex and sex hormones on human chronic disorders of complex aetiology), a 2-­year project ORCID iDs (2017–2019) coordinated by Hospital Ramón y Cajal Jesus Lopez-­Alcalde http://​orcid.​org/​0000-​0002-​1598-​8790 Alba Antequera Martín http://​orcid.​org/​0000-​0002-​1670-​6302 (Madrid, Spain). We shall disseminate our findings Alfonso Muriel http://​orcid.​org/​0000-​0002-​4805-​4011 through peer-­reviewed publication, at conferences, and Ivan Solà http://​orcid.​org/​0000-​0003-​0078-​3706 in the mass media. Rosa del Campo http://​orcid.​org/​0000-​0003-​1147-​7923 4 Lopez-­Alcalde J, et al. BMJ Open 2020;10:e035927. doi:10.1136/bmjopen-2019-035927 Open access Manuel Ponce-­Alonso http://​orcid.​org/​0000-​0002-​3239-​9373 24 Paul M, Shani V, Muchtar E, et al. Systematic review and meta-­ Noelia Álvarez-­Díaz http://​orcid.​org/​0000-​0002-​5904-​3609 analysis of the efficacy of appropriate empiric antibiotic therapy for Olaya Madrid-­Pascual http://​orcid.​org/​0000-​0002-​2861-​0121 sepsis. Antimicrob Agents Chemother 2010;54:4851–63. 25 Gaieski DF, Mikkelsen ME, Band RA, et al. Impact of time to Javier Zamora http://​orcid.​org/​0000-​0003-​4901-​588X antibiotics on survival in patients with severe sepsis or septic shock in whom early goal-­directed therapy was initiated in the emergency department. Crit Care Med 2010;38:1045–53. 26 Ferrer R, Martin-­Loeches I, Phillips G, et al. Empiric antibiotic References treatment reduces mortality in severe sepsis and septic shock 1 Singer M, Deutschman CS, Seymour CW, et al. The third from the first hour: results from a guideline-­based performance International consensus definitions for sepsis and septic shock improvement program. Crit Care Med 2014;42:1749–55. (Sepsis-3). JAMA 2016;315:801. 27 Wilczynski NL, Haynes RB, Hedges Team. Developing optimal 2 Linde-­Zwirble WT, Angus DC. Severe sepsis epidemiology: sampling, search strategies for detecting clinically sound prognostic studies in selection, and society. Crit Care 2004;8:222–6. MEDLINE: an analytic survey. BMC Med 2004;2:23. 3 Perner A, Gordon AC, De Backer D, et al. Sepsis: frontiers in 28 Lorenzetti DL, Lin Y. Locating sex- and gender-­specific data in diagnosis, resuscitation and antibiotic therapy. Intensive Care Med health promotion research: evaluating the sensitivity and precision of 2016;42:1958–69. published filters. J Med Libr Assoc 2017;105:216-225. 4 Fleischmann C, Scherag A, Adhikari NKJ, et al. Assessment of global 29 Moerman CJ, Deurenberg R, Haafkens JA. Locating sex-­specific incidence and mortality of Hospital-­treated sepsis. current estimates evidence on clinical questions in MEDLINE: a search filter for use on and limitations. Am J Respir Crit Care Med 2016;193:259–72. OvidSP. BMC Med Res Methodol 2009;9:25. 5 Reinhart K, Daniels R, Kissoon N, et al. Recognizing Sepsis 30 Song MM, Simonsen CK, Wilson JD, et al. Development of a as a Global Health Priority - A WHO Resolution. N Engl J Med PubMed based search tool for identifying sex and gender specific 2017;377:414–7. health literature. J Womens Health 2016;25:181–7. 6 Moskowitz A, Omar Y, Chase M, et al. Reasons for death in patients 31 Stewart F, Fraser C, Robertson C, et al. Are men difficult to find? with sepsis and septic shock. J Crit Care 2017;38:284–8. identifying male-­specific studies in MEDLINE and Embase. Syst Rev 7 Pinheiro da Silva F, César Machado MC. Personalized medicine for 2014;3:78. sepsis. Am J Med Sci 2015;350:409–13. 32 Endnote X7. Endnote X7 computer program. Version X 7.8. 8 Esper AM, Moss M, Lewis CA, et al. The role of infection and Philadelphia: Clarivate Analytics, 2018. comorbidity: factors that influence disparities in sepsis. Crit Care 33 Thomas J, Brunton JGS. EPPI-­Reviewer 4: software for research Med 2006;34:2576–82. synthesis, 2010. 9 Martin GS, Mannino DM, Eaton S, et al. The epidemiology of 34 Polanin JR, Pigott TD, Espelage DL, et al. Best practice guidelines sepsis in the United States from 1979 through 2000. N Engl J Med for Abstract screening large‐evidence systematic reviews and meta‐ 2003;348:1546–54. analyses. Res Synth Methods 2019;10:330–42. 10 Tsertsvadze A, Royle P, Seedat F, et al. Community-­Onset sepsis and 35 Riley RD, Moons KGM, Snell KIE, et al. A guide to systematic its public health burden: a systematic review. Syst Rev 2016;5:81. review and meta-­analysis of prognostic factor studies. BMJ 11 Papathanassoglou E, Middleton N, Benbenishty J, et al. Systematic 2019;364:k4597. review of gender- dependent outcomes in sepsis. Nurs Crit Care 36 Moons KGM, de Groot JAH, Bouwmeester W, et al. Critical appraisal 2017;22:284–92. and data extraction for systematic reviews of prediction modelling 12 Sakr Y, Elia C, Mascia L, et al. The influence of gender on the studies: the charms checklist. PLoS Med 2014;11:e1001744. epidemiology of and outcome from severe sepsis. Crit Care 37 Hayden JA, van der Windt DA, Cartwright JL, et al. Assessing bias in 2013;17:R50. studies of prognostic factors. Ann Intern Med 2013;158:280. 13 Eachempati SR, Hydo L, Barie P. Bending gender rules for septic 38 Aldin A, Umlauff L, Estcourt LJ, et al. Interim PET-­results for patients: are host responses positioned equally for all critically ill prognosis in adults with Hodgkin lymphoma: a systematic review and patients? Crit Care Med 2009;37:2649–50. meta-­analysis of prognostic factor studies. Cochrane Database Syst 14 Croce MA, Fabian TC, Malhotra AK, et al. Does gender difference Rev 2019;9:CD012643. influence outcome? J Trauma 2002;53:889–94. 39 Symons MJ, Moore DT. Hazard rate ratio and prospective 15 Riley RD, Hayden JA, Steyerberg EW, et al. Prognosis research epidemiological studies. J Clin Epidemiol 2002;55:893–9. strategy (progress) 2: prognostic factor research. PLoS Med 40 The Nordic Cochrane Centre. Review Manager (RevMan) [Computer 2013;10:e1001380. program. Version 5.3. Copenhagen: The Nordic Cochrane Centre, 16 Shamseer L, Moher D, Clarke M, et al. Preferred reporting items for The Cochrane Collaboration, 2014. systematic review and meta-­analysis protocols (PRISMA-­P) 2015: 41 Borenstein M, Higgins JPT, Hedges LV, et al. Basics of meta-­analysis: elaboration and explanation. BMJ 2015;349:g7647. I 2 is not an absolute measure of heterogeneity. Res Synth Methods 17 Moher D, Liberati A, Tetzlaff J, et al. Preferred reporting items for 2017;8:5–18. systematic reviews and meta-­analyses: the PRISMA statement. PLoS 42 Borenstein M. Common Mistakes in Meta-­Analysis and How to Avoid Med 2009;6:e1000097. Them. First edn. Englewood: Biostat, Inc, 2019. 18 Johnson JL, Greaves L, Repta R. Better science with sex and 43 Peters JL, Sutton AJ, Jones DR, et al. Comparison of two methods gender: facilitating the use of a sex and gender-­based analysis in to detect publication bias in meta-­analysis. JAMA 2006;295:676. health research. Int J Equity Health 2009;8:14. 44 Guyatt G, Oxman AD, Akl EA, et al. Grade guidelines: 1. Introduction-­ 19 Vincent J-­L, Moreno R, Takala J, et al. The SOFA (sepsis-­related GRADE evidence profiles and summary of findings tables. J Clin organ failure assessment) score to describe organ dysfunction/ Epidemiol 2011;64:383–94. failure. Intensive Care Med 1996;22:707–10. 45 Schünemann H, Brozek J, et al, The GRADE Working Group. 20 Tolsma V, Schwebel C, Azoulay E, et al. Sepsis severe or septic Grade Handbook for grading quality of evidence and strength of shock: outcome according to immune status and immunodeficiency recommendations, 2019. ​fromgdt.​guidelinedevelopment.​org/​app/​ profile. Chest 2014;146:1205–13. handbook/​handbook.​html 21 Danai PA, Moss M, Mannino DM, et al. The epidemiology of sepsis in 46 Huguet A, Hayden JA, Stinson J, et al. Judging the quality of patients with malignancy. Chest 2006;129:1432–40. evidence in reviews of prognostic factor research: adapting the grade 22 O'Brien JM, Lu B, Ali NA, et al. Alcohol dependence is independently framework. Syst Rev 2013;2:71. associated with sepsis, septic shock, and hospital mortality among 47 Iorio A, Spencer FA, Falavigna M, et al. Use of grade for assessment adult intensive care unit patients. Crit Care Med 2007;35:345–50. of evidence about prognosis: rating confidence in estimates of event 23 Krieger JN, Kaiser DL, Wenzel RP. Urinary tract etiology of rates in broad categories of patients. BMJ 2015;350:h870. bloodstream infections in hospitalized patients. J Infect Dis 48 GRADEpro GDT, McMaster University. Guideline development tool 1983;148:57–62. [computer program. Evidence Prime, Inc, 2015. http://​grade.​pro.​org/ Lopez-­Alcalde J, et al. BMJ Open 2020;10:e035927. doi:10.1136/bmjopen-2019-035927 5