Global prevalence and determinants of hospital-acquired COVID-19: protocol for a systematic review and meta-analysis Authors: Noach Leon Ribaric, Zixue O’Neill-Byrne, Achim Hellinger, Daniel Pan, Nick M Wilson, Ignazio Maria Viola, Timothy S Walsh, Tom Lawton, Sergejs Afanasjevs, Clementina Calabria, Federico De Angelis, Thomas Theo Brehm, Antonia Zapf Published: 2026 ABSTRACT Abstract AbstractIntroductionA proportion of patients hospitalised for COVID-19 acquire the disease during their hospital stay, underscoring the risk of hospital-acquired COVID-19 (HA-COVID-19). This risk is presumed to be high, given how commonly and intensely air and surfaces within hospitals are reportedly contaminated with SARS-CoV-2. However, the true extent of HA-COVID-19 worldwide remains unknown, with limited understanding of factors that influence its occurrence and how these have evolved over time. This review will therefore aim to estimate the pooled prevalence of HA-COVID-19 among hospitalised COVID-19 patients globally and investigate differences by country, type of hospitals, medical specialty, length and timing of studied periods.Methods and analysisA systematic review and meta-analysis will be conducted adhering to PRISMA (Preferred Reporting Items for Systematic Reviews and Meta-Analyses) guidelines. MEDLINE and PubMed Central (via PubMed), Scopus, Embase (via Ovid), the Web of Science Core Collection as well as websites of public health agencies (PHA) will be searched until 1 July 2026. All journal articles and sources from PHAs reporting any primary data on the prevalence of HA-COVID-19 will be included. Methodological quality will be assessed using the Joanna Briggs Institute Critical Appraisal Checklist for Studies Reporting Prevalence Data. The primary outcome will be the global prevalence of HA-COVID-19. Data synthesis will include random-effects proportional meta-analysis. Estimates will be presented with two-sided 95% CIs and heterogeneity assessed using the I² statistic.Ethics and disseminationEthical approval is not needed as no original data will be generated. This review will be published in an international, peer-reviewed journal.PROSPERO registration numberCRD420251136884.Keywords: COVID-19, Systematic Review, Meta-Analysis, SARS-CoV-2 Infection, Infection control, PUBLIC HEALTH STRENGTHS AND LIMITATIONS OF THIS STUDY This review will make use of a multidisciplinary developed and piloted search strategy to identify peer-reviewed journal articles. This review will make use of a multidisciplinary developed and piloted search strategy to identify peer-reviewed journal articles. It will additionally incorporate data from public health agencies for a more comprehensive and representative overview. It will additionally incorporate data from public health agencies for a more comprehensive and representative overview. Variability in the definitions of hospital-acquired COVID-19 across studies may pose challenges to data analysis. Variability in the definitions of hospital-acquired COVID-19 across studies may pose challenges to data analysis. Differences in methodological approaches and outcome measures in studies on hospital-acquired COVID-19 may increase the number of studies excluded from this review. Differences in methodological approaches and outcome measures in studies on hospital-acquired COVID-19 may increase the number of studies excluded from this review. Introduction The COVID-19 pandemic has posed an unprecedented challenge to infection prevention and control (IPC) in hospitals worldwide. As of 2025, over 700 million individuals have been infected with SARS-CoV-2,1 and it is estimated that about 5% needed to be hospitalised.2 A proportion of these patients are believed to have acquired the infection during their hospital stay, making nosocomial transmission a notable contributor to the pandemic’s overall burden on society.3 Although several studies have reported how many patients had hospital-acquired COVID-19 (HA-COVID-19),4,17 a comprehensive overview of its global prevalence remains lacking. This is because many of them focused on periods as short as a few weeks,4,10 or on as few as a single hospital,11,17 limiting the extent to which their results can be generalised. Reported rates and covered countries have also varied markedly,18 with much of this heterogeneity remaining unexplained. It has been hypothesised that differences in pandemic waves, public health infrastructure across countries, infection control practices within hospitals, approaches to testing for SARS-CoV-2 and investigated medical specialties may help explain this variation.19 However, this combination of infrequent estimates of HA-COVID-19, data with limited generalisability and largely unexplained heterogeneity currently restricts the ability to develop practical recommendations for IPC. Further considering the ongoing circulation of COVID-19 in the community20 and its well-established interplay with rates of hospital-acquired infections (HAIs),21 an improved understanding of the global extent and drivers of HA-COVID-19 is urgently needed, if healthcare systems intend to better protect the vulnerable population of hospitalised patients from preventable harm in the future.22 Study objectives This systematic review and meta-analysis will aim to estimate the pooled prevalence of HA-COVID-19 among hospitalised adult patients infected with SARS-CoV-2 globally. Due to expected variation in prevalences, these will further be stratified by country; type and number of investigated hospitals; type of data source; type of medical specialty; length of investigated period; methodology and frequency of testing; type of practiced precautions against and definitions of HA-COVID-19. Finally, it will be identified how the emergence of Omicron variants modified the prevalence of HA-COVID-19 to evaluate the effect of pathogen evolution over time. Methods Protocol design and registration This study protocol was established in alignment with the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) reporting guidelines for Protocols (PRISMA-P),23 see the online supplemental appendix 1. To promote methodological transparency and research integrity, it was also registered in the Prospective Register of Systematic Reviews (PROSPERO),24 under the registration ID: CRD420251136884. Eligibility criteria All peer-reviewed journal articles and data sources from public health agencies (PHAs) providing primary data on the prevalence of HA-COVID-19 in adult hospitalised patients infected with SARS-CoV-2, or from which prevalences can be calculated, will be included in this systematic review and meta-analysis. All study designs, including deposited datasets, will be included, apart from qualitative study designs not reporting prevalence data, modelling or animal studies, analyses of claims data, case reports/case series, outbreak analyses or narrative reviews. Secondary analyses in the form of systematic reviews will be assessed by full-text review only to retrieve reviewed original articles that might have been missed by database search, cross-referencing or snowballing, that is, citation chaining. Articles reporting data on paediatric inpatients, numbers of community-acquired COVID-19 (CA-COVID-19) exclusively, other healthcare settings apart from hospitals, occupationally-acquired COVID-19, only non-COVID-19 HAIs, HA-COVID-19 without CA-COVID-19 or incidences of HA-COVID-19 will also be excluded. Studies reporting data on both HA-COVID-19 and other HAIs will be included, but only HA-COVID-19 data will be analysed. Analyses will similarly be conducted when studies examine both adult and paediatric inpatients. The definition of what constitutes HA-COVID-19 will follow that of the original sources. When no such determination is made, the HA-COVID-19 definition used in previous research will be employed (table 1).25 Table 1Definitions of community-acquired COVID-19 and hospital-acquired COVID-19Dichotomous codingCategoryDefinitionCA.COVID.19Community-onset community-acquired (CO.CA)Positive test date <14 days pre-admission/attendance and up to day 2 of admission; no prior discharge within 14 days of admission/attendanceCommunity-onset possible healthcare-associated (CO.pHA)Positive test date ≤14 days post-discharge; if readmitted during this period, up to day 2 of admission where date of readmission is day 1Hospital-onset indeterminate healthcare-associated (HO.iHA)Positive test from day 3 to day 7 of admission, inclusivelyUnclassified (U)All cases which do not meet any of the following criteriaHA.COVID.19Hospital-onset probable healthcare-associated (HO.pHA)Positive test from day 8 to day 14 of admission, inclusivelyHospital-onset definite healthcare-associated (HO.HA)Positive test from day 15 of admission until day of discharge, inclusively Definitions of community-acquired COVID-19 and hospital-acquired COVID-19 If prevalences of HA-COVID-19 (%HA.COVID.19) need to be calculated from data, the following equation will be used: where HA.COVID.19 and CA.COVID.19 are the numbers of HA-COVID-19 and CA-COVID-19, respectively. Information sources Systematic literature searches, cross-referencing and snowballing of articles will be performed within the online databases of MEDLINE and PubMed Central (via PubMed), Scopus, Embase (via Ovid) and the Web of Science Core Collection, published at any time up until 1 July 2026. Results will be limited to peer-reviewed journal articles published in languages the authors are fluent in: English, German, Spanish, Italian and Croatian. In addition, websites of PHAs will be searched and, if needed, departments of PHAs contacted for sources providing original data fulfilling the eligibility criteria laid out above. Search strategy To ensure that all relevant articles will be identified, the PRISMA extension for search (PRISMA-S),26 the Peer Review of Electronic Search Strategies Evidence-Based Checklist (PRESS EBC)27 as well as librarians from the University Medical Centre Hamburg-Eppendorf’s Central Medical Library were consulted when developing search strings. Where possible (in MEDLINE and Embase), titles, abstracts, keywords and full texts will be searched and, in Scopus and the Web of Science Core Collection, only titles, abstracts and keywords will be retrieved using a combination of boolean operators and suitable search terms: (SARS-CoV-2 OR COVID-19) AND (hospital OR hospitals OR in-hospital OR intrahospital) AND (nosocomial OR hospital-acquired OR hospital-onset OR hospital-associated OR healthcare-acquired OR healthcare-associated OR healthcare-onset). Selection process Articles will be screened by title and duplicates subsequently excluded. Remaining unique citations will be assessed by review of abstracts. Article selection for and full-text review of studies will be performed by two authors (NLR, ZO’N-B) working independently while adhering to the identification and screening guidelines established by the 2020 PRISMA statement.28 Any disparities between authors will be resolved by discussion. Data items As per the data extraction form (online supplemental appendix 2), the following data will be manually extracted from included studies: author names; journal title; article language and database it was identified with; publishing date; digital object identifier or link; country; WHO region; number, location and type of studied hospitals; type of studied medical specialties; type of data source; study design; start and end date of the investigation; number of patients with HA-COVID-19 (n), number of patients with CA-COVID-19, total number of hospitalised COVID-19 patients (N) and prevalence of HA-COVID-19 (%); methodology and frequency of testing; type of practiced precautions against and definition of HA-COVID-19. To allow for multiple comparisons of study estimates, several rows per study will be created, if, for example, Omicron and pre-Omicron periods were investigated or several definitions of HA-COVID-19 were explored. In case weekly or monthly numbers of HA-COVID-19 are available, these will be extracted separately. Data management and data collection process Two authors (NLR, ZO’N-B) will extract data independently. If discrepancies in data handling arise, authors will discuss and resolve any issues. Missing data in included studies will be treated as missing completely at random and authors of primary studies will not be contacted. Outcomes and prioritisation The primary outcome of this systematic review and meta-analysis is the global prevalence of HA-COVID-19 among hospitalised, adult patients with a SARS-CoV-2 infection. Differences in prevalences will be explored in relation to country or WHO region, type and number of hospitals, type of data source and medical specialty, methodology and frequency of testing, type of practiced precautions, length and timing of studied period and definitions of HA-COVID-19. Risk of bias in individual studies During data extraction, two authors (NLR, ZO’N-B) will independently assess risks of bias in studies using the Joanna Briggs Institute (JBI) Critical Appraisal Checklist for prevalence studies,29 as suggested by a recent systematic review on that matter.30 JBI Quality Scores will be synthesised in tabular form and disagreements about them will be resolved by discussion. No sources will be excluded from the review based on this assessment; however one of three sensitivity analyses in this review will explore whether prevalence estimates change when studies with the highest risk of bias were to be excluded. Data synthesis A random-effects, proportional meta-analysis of the data will be conducted with R’s meta and metafor packages to estimate the global, pooled prevalence of HA-COVID-19 in hospitalised, adult patients with a SARS-CoV-2 infection. In preparation, raw prevalences will be logit-transformed to stabilise variances.31 For studies with 0% prevalence, the Haldane-Anscombe correction will be applied.32 33 Results will be presented with 95% CIs and visualised in a forest plot, both for the entirety of analysed periods and separately for pre-Omicron and Omicron waves. Heterogeneity across studies will be evaluated with Cochran’s Q, Tau2 and the I2 statistic.34 Where possible, subgroup analyses will be performed in regards to the confounders listed in ‘Outcome and prioritisation’. Subgroup differences will be assessed using χ2 tests. Additionally, simple and multi-level, multiple meta-regression analyses will be computed to identify covariates responsible for the observed variation. The hierarchical structure within multi-level multiple regressions will comprise effect sizes at level 1 and study-level clustering at level 2, with study-level covariates included as fixed-effect moderators. As done for the pooled prevalence analysis, heterogeneity in meta-regressions will be determined using Tau2 and the I2 statistic.34 Predictor effects and unexplained heterogeneity will be evaluated using the omnibus test for moderators (QM) and test for residual heterogeneity (QE). The contributions of predictors to intra-study and inter-study heterogeneity will be calculated with pseudo-R2. Regression coefficients will be tabulated and forest plots created. A zero-model including only the intercept will be initially fitted. Variable selection for inclusion in multiple regression analyses will then be guided by biological plausibility and mutual exclusivity to avoid collinearity and minimise residual heterogeneity. Model performance will be evaluated with the Akaike information criterion, Bayesian information criterion, QM, QE and pseudo-R2. If weekly or monthly rates of HA-COVID-19 are further available, these will be displayed graphically. Meta-biases The risk of bias across studies will be ascertained by evaluating publication bias with the Egger test and the funnel plot’s symmetry. Presence of bias will be assumed in case of an asymmetric funnel plot or if it is indicated by the Egger test (p<0.2).35 The second sensitivity analysis of this review will further estimate each study’s contribution to the pooled prevalence of HA-COVID-19 worldwide. The last one will compare estimates between small and large-scale studies. Confidence in cumulative evidence The overall quality of evidence is commonly assessed using the Grading of Recommendations Assessment, Development and Evaluation (GRADE) approach, which considers the domains of risk of bias, imprecision, inconsistency, indirectness and publication bias.36 GRADE was originally developed for evaluating outcomes in randomised controlled trials and its emphasis on criteria such as randomisation and blinding limits its applicability to observational study designs.36 Observational studies, however, are most suitable for estimating real-world prevalences and are expected to constitute the majority of included studies in this systematic review and meta-analysis. Applying GRADE to these studies, would lead to an unjustified downgrade in quality due to the inherent confounding and selection biases of observational research, despite the studies providing valid outcomes. As a result, the GRADE approach will not be employed. In the absence of a validated alternative framework for assessing the quality of evidence in prevalence studies, the overall quality of evidence will not be determined here. Patient and public involvement Patients or the public were not involved in the design, conduct, reporting or dissemination plans of this research protocol. Ethics and dissemination No ethical approval is required for this review. A manuscript containing the results of this research will be submitted to an international, peer-reviewed journal. Findings will also be disseminated through international academic meetings and conferences and will be used for further research. Protocol deviations The presence of and reasons for any meaningful deviations from this protocol will be reported in the final manuscript. Contribution to the literature A systematic search for previously published reviews on this topic identified 10 papers summarising evidence from multiple hospital settings.1837,45 Unfortunately, these were either of narrative design,37 38 limited to one medical specialty,39 40 focused exclusively on retrospective hospital outbreak analyses and genomic surveillance studies41 or covered the literature of the beginning of the COVID-19 pandemic only,1842,45 with search dates ranging from 31 March 202042 to 03 May 2020,43 30 November 2020,18 09 February 202144 and 01 June 2021,45 respectively. As such, it is critical to provide a systematic, updated, comprehensive and quantitative synthesis of the prevalence of HA-COVID-19. The planned systematic review and meta-analysis will, for the first time, evaluate the pooled prevalence of HA-COVID-19 globally and throughout all pandemic phases. It is conceptually unique in its explicit focus on the drivers of HA-COVID-19 and their relationships to each other. It also differs methodologically from conventional meta-analyses by not only retrieving peer-reviewed studies but also incorporating raw data and reports of PHA to provide a more representative overview. As for publishing any research protocol, this protocol outlined here will help prevent duplication, foster transparency, minimise the risk of selective reporting and enhance reproducibility.46 These aspects are particularly important for HA-COVID-19 given the known heterogeneity across studies in terms of settings, practiced precautions, pandemic phases, reporting standards and definitions of HA-COVID-19.4,17 Supplementary material Acknowledgements We would like to thank Erin McGill and Diane Lee (both from the Public Health Agency of Canada) very much for their support and the Canadian Nosocomial Infection Surveillance Program (CNISP) for collecting and sharing primary data on hospital-acquired COVID-19. Footnotes 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. Prepublication history and additional supplemental material for this paper are available online. To view these files, please visit the journal online (https://doi.org/10.1136/bmjopen-2025-113205). Provenance and peer review: Not commissioned; externally peer reviewed. Patient consent for publication: Not applicable. Patient and public involvement: Patients and/or the public were not involved in the design, or conduct, or reporting, or dissemination plans of this research. References Supplementary Materials