Introduction to hantaviruses

Haemorrhagic fever with renal syndrome (HFRS) is a zoonotic disease caused by several rodent-borne viruses of the genus Orthohantavirus in Europe and Asia. Hantaviruses are members of the family Hantaviridae, order Bunyavirales, class Ellioviricetes [1]. The family Hantaviridae is divided into multiple subfamilies and genera that infect a wide range of hosts, including rodents, shrews, moles, bats, reptiles and fish, but all known human pathogens belong to the genus Orthohantavirus [2]. Members of this genus are associated with persistent infection in specific rodent reservoirs, from which they are transmitted to humans primarily through the inhalation of aerosolised excreta (urine, faeces, saliva), or less commonly by rodent bites [3,4]. Virions are enveloped, negative-sense, single-stranded RNA viruses with a tri-segmented genome [5]. The order Bunyavirales is large and includes several other genera of major public health importance beyond Hantaviridae, such as Nairoviridae (Crimean–Congo haemorrhagic fever virus), Arenaviridae (Lassa virus), and Phenuiviridae (Rift Valley fever virus, severe fever with thrombocytopenia syndrome virus) [6], all of which have epidemic potential and cause severe disease in humans.

In Europe and Asia, hantaviruses cause HFRS, whereas in the Americas the clinical syndrome is hantavirus cardiopulmonary syndrome (HCPS) [3]. Across Europe, reported HFRS notifications fluctuate substantially from year to year: between 2019 and 2023, annual rates ranged from 0.4 to 1.1 per 100,000 population, with 1,885 reported cases in 2023 and 4,046 cases in 2019, reflecting typical interannual variability driven by rodent population dynamics [7]. In 2023, Finland and Germany together accounted for 60.5% of European reports, and endemic activity is also well documented in Sweden, Belgium, Norway, France, Austria and Slovenia [7,8]. In Asia, China carries the heaviest burden globally, with national surveillance consistently recording tens of thousands of HFRS cases each year, estimated at roughly 20,000–50,000 annually in recent decades [9,10]. South Korea also reports sustained transmission, with several hundred cases reported annually and clear seasonal peaks [11]. By contrast, across North and South America combined, only a few hundred HCPS cases are typically reported each year [12]. It is likely that the true global incidence is higher, as asymptomatic or mild infections often go undetected [3].

Four Orthohantavirus species are recognised as the main causes of HFRS in Europe and Asia, each closely associated with a primary rodent reservoir (Table 1) [13]. In Europe, Orthohantavirus puumalaense (Puumala virus, PUUV) accounts for the majority of cases, typically causing the milder disease. Orthohantavirus dobravaense (Dobrava virus, DOBV) occurs mainly in the Balkans and parts of central and eastern Europe, where it is associated with more severe disease but fewer cases overall [14]. In Asia, Orthohantavirus hantanense (Hantaan virus, HTNV) is an important cause of severe disease, particularly in rural areas, whereas Orthohantavirus seoulense (Seoul virus, SEOV) is associated with moderate disease and has a wider distribution, maintained in commensal rats in both urban and peri-urban settings [15]. These four viruses represent the principal human pathogens, although other related viruses within the same species complexes have also been shown to infect humans and are summarised in Table 1. In addition, a fifth species, Orthohantavirus tulaense (Tula virus, TULV), has occasionally been reported in human infection, but cases are rare and typically mild, and its pathogenic significance remains uncertain [5].

Download:

• PNG
  larger image
• TIFF
  original image

Table 1. ICTV-recognised Orthohantavirus species causing HFRS in Europe and Asia.

https://doi.org/10.1371/journal.pntd.0014524.t001

The emergence of HFRS in humans is closely tied to rodent populations, which fluctuate with climate and human activity [12]. An increase in the frequency and magnitude of HFRS epidemics in Europe has been partially attributed to the warming climate and increased land-use [16,17]. The epidemic potential of HFRS could increase further with a warming climate, possibly leading to more frequent outbreaks in endemic areas, and emergence of cases in new regions.

Eurasian hantaviruses infect microvascular endothelial cells and can disseminate to infect endothelial cells in most major organs. There is, however, prominent involvement of the renal microvasculature, particularly glomerular and tubular endothelial cells, contributing to the characteristic renal manifestations of disease [18]. Infected endothelial cells exhibit features of endothelial cell activation, with increased microvascular permeability and leakage of intravascular fluid into the extracellular compartment. Hantaviruses also trigger a vigorous immune response that contributes to pathological degrees of intravascular fluid leakage [19,20]. Varying degrees of acute renal failure occur secondary to this pathological process.

Acute thrombocytopenia is another central pathogenic feature in HFRS, with evidence of increased platelet turnover, hypothesised to be secondary to widespread adherence of circulating platelets to infected endothelium [21]. In keeping with this, bone marrow shows evidence of increased production with elevated thrombopoietin levels found in some studies [22,23]. Infection of the endothelium causing endothelial cell activation will result in a localised prothrombotic state. This, however, does not explain the bleeding tendency seen [24]. Some studies have demonstrated evidence of platelet dysfunction, but whether this is important in the development of haemorrhagic manifestations is poorly understood [21,23,25].

After an incubation period of ten days to four weeks, there is an abrupt onset of fever, typically associated with headache, myalgia, back pain and vomiting [3]. HFRS is classically described as occurring in five phases: febrile, hypotensive, oliguric, polyuric, and convalescent. Haemorrhagic manifestations typically occur during the hypotensive and oliguric phases of infection. These range from minor bleeding from small blood vessels in the skin, eyes and mouth, to major haemorrhage in the gastrointestinal tract, urinary tract and the brain [26]. HFRS can range from subclinical seroconversion to rapidly progressive and fatal. PUUV infection, known as nephropathia epidemica (NE) in Europe, is generally mild with a low case fatality rate (CFR), typically <0.1%, although estimates from Sweden suggest rates up to 0.4% [12,27]. Mild haemorrhagic symptoms are observed in around one third of NE cases [3]. In contrast, DOBV typically leads to a severe disease course in Europe with more significant bleeding, haemodynamic instability and renal failure, with case fatality rates estimated to be around 10% [28]. HTNV results in similarly severe disease in Asia with CFR estimates between 5 – 15% [3]. SEOV causes moderate disease severity in Asia and a CFR of <1 – 2% [28]. Nevertheless, cases of PUUV infection can be fulminant and fatal and DOBV or HTNV can result in mild disease.

Treatment of HFRS is entirely supportive. Haemodialysis use in severe renal failure, and transfusion of blood products in patients with severe thrombocytopenia or bleeding may be required. No licensed therapeutic agents currently exist for use in HFRS [3]. Diagnosis is dependent on a combination of clinico-epidemiological features and laboratory tests. Enzyme-linked immunosorbent assay (ELISA), indirect immunofluorescence assay (IFA) and reverse transcription polymerase chain reaction (RT-PCR) are widely used methods [3]. {Vial, 2023 #138}{Vial, 2023 #138}{Vial, 2023 #138}{Vial, 2023 #138}{Avsic-Zupanc, 2019 #103}{Avsic-Zupanc, 2019 #103}

Haemostasis in HFRS

Microvascular endothelial dysfunction and/or activation induced by hantavirus infection is a cardinal pathological feature in HFRS and is likely a critical component in the prothrombotic change [24]. Separately, infected microvasculature may be predisposed to rupture giving rise to minor skin and mucosal petechiae, which would be exacerbated by thrombocytopenia [24]. Endothelial activation and injury in the early phases of hantavirus infection is a possible trigger for additional pathological coagulation disturbances as upregulation of tissue factor and activation of Factor XII have been described [29,30].

Indeed, it is well recognised that endothelial cell activation and injury expose subendothelial tissue factor, triggering the extrinsic coagulation pathway [24]. Hantavirus infection has also been shown to activate the intrinsic clotting pathway [30]. Both pathways converge at the activation of Factor X, which then converts prothrombin to thrombin, the key enzyme that transforms fibrinogen into insoluble fibrin strands forming the structural basis of the clot [31]. Widespread activation of the clotting cascade in hantavirus infection could cause a consumptive coagulopathy, indeed a disseminated intravascular coagulation (DIC), resulting in depletion of coagulation factors and diminished thrombin generation capacity. However, severe prolongation of PT and APTT reflecting significant loss of coagulation factors is rarely observed in HFRS, suggesting that this mechanism is unlikely to represent the primary pathway of haemostatic dysfunction [24]. PUUV-infected patients have also previously been shown to demonstrate enhanced fibrinolytic potential due to increased levels of tissue-type plasminogen activator (t-PA), which is sometimes a feature of endothelial cell activation [32]. It has been hypothesised that a combination of thrombocytopenia, diminished thrombin generation capacity, and hyperfibrinolysis shift the haemostatic balance towards hypocoagulability and haemorrhagic tendency [24].

Fully understanding the mechanisms through which hypocoagulability occurs in HFRS would provide important targets for novel therapeutic agents and alter clinical management strategies. Unfortunately, routine laboratory clotting parameters provide little insight into the causes of an individual patient’s haemostatic balance.

This systematic review aimed to characterise the laboratory clotting abnormalities and haemorrhagic manifestations observed in HFRS and explore differences between hantavirus species. These aims were achieved through a comprehensive synthesis of over 7,000 published cases, enabling detailed comparison between PUUV, DOBV, HTNV and SEOV. Notably, this is the first review to systematically quantify the frequency of specific haemorrhagic manifestations and compare them across different hantaviruses, highlighting under-recognised patterns of bleeding and key gaps in current reporting practices.

Database searches

This systematic review was conducted in accordance with Preferred Reporting Items for Systematic reviews and Meta-Analyses (PRISMA) guidelines [33]. Database searches were conducted on 12 December 2024 using MEDLINE, PubMed, CINAHL, Web of Science and Scopus. The search terms used are outlined in Table 2. No date restrictions were applied to the database searches. This review was registered on PROSPERO on 25^th November 2024 (CRD42024618760), and the protocol is publicly available at https://www.crd.york.ac.uk/PROSPERO/view/CRD42024618760.

Download:

• PNG
  larger image
• TIFF
  original image

Table 2. Systematic review database search terms.

https://doi.org/10.1371/journal.pntd.0014524.t002

Study selection

To be included in this review, studies had to meet all the following criteria: (1) observational study; (2) any human infection with hantaviruses causing HFRS; (3) laboratory confirmed infection using serological methods for specific IgM and/or IgG detection OR viral RNA detection using RT-PCR; (4) data reported on laboratory clotting parameters and/or haemorrhagic manifestations. Studies with any of the following were excluded: (1) suspected or confirmed infection with other pathogens; (2) reporting only severe or fatal cases of HFRS; (3) sample size <10 patients; (4) any study conducted at the same hospital site with an overlapping recruitment period to another included study; (5) studies with an interventional component without any baseline clinical data reported prior to intervention.

All references retrieved from the literature search were imported into Rayyan, a web-based tool designed to facilitate systematic reviews. Rayyan’s duplicate detection feature was employed to identify and remove duplicate records prior to screening. Two reviewers (MR and AH) independently screened titles and abstracts against the predefined inclusion criteria within Rayyan. For studies deemed potentially relevant, full-text articles were assessed independently by the same reviewers. Rayyan’s conflict detection functionality highlighted any discrepancies between reviewers’ decisions. These conflicts were initially resolved through discussion between the two reviewers; if consensus could not be reached, a third reviewer (TF) adjudicated the decisions.

Data extraction

The process through which publications were identified, screened and included is depicted in Fig 1. A total of 55 studies were finally included in this review. Each publication was read in-depth. Demographic, laboratory and clinical data were extracted and collated in an excel spreadsheet. The laboratory parameters of interest were: platelet count, prothrombin time (PT), activated partial thromboplastin time (APTT), aspartate aminotransferase (AST), alanine aminotransferase (ALT), fibrinogen, haemoglobin, haematocrit, and d-dimer.

Download:

• PNG
  larger image
• TIFF
  original image

Fig 1. PRISMA flow diagram.

Flowchart outlines the process from database searches through to final study inclusion. “n=” denotes the number of publications retrieved, excluded or included at each stage.

https://doi.org/10.1371/journal.pntd.0014524.g001

Data extraction was primarily conducted by MR using a standardised Excel spreadsheet and AH independently verified the accuracy of the extracted data. Any discrepancies identified were first discussed between the two reviewers to determine if they resulted from omissions or transcription errors and were corrected accordingly. If disagreements persisted, TF acted as a third reviewer to make a final decision.

The following data were extracted from each included study:

• Study details: study design, sample size, recruitment period, and location (country, city, hospital(s))
• Demographic details: age and gender of participants
• Causative hantavirus: specific virus identified as the cause
• Laboratory values: as specified above
• Haemorrhagic manifestations: number observed, site, and severity
• Clinical outcome: survival versus death

Haemorrhagic manifestations refer to clinically reported bleeding events described in the included studies (e.g., petechiae, mucosal bleeding, haematuria, gastrointestinal bleeding, pulmonary haemorrhage, intracranial haemorrhage). Microscopic haematuria refers to blood in the urine detectable only on laboratory testing (e.g., urine microscopy or dipstick), whereas macroscopic haematuria refers to visibly blood-stained urine. Patients with laboratory abnormalities alone were not recorded as having haemorrhagic manifestations unless clinically evident or radiologically confirmed bleeding was documented.

Instances of missing data were recorded as such during the data extraction process. These cases were initially discussed between MR and AH to assess their potential impact on the outcomes of the review. As the review progressed, it became evident that a meta-analysis would not be feasible due to the heterogeneity of reporting and the frequent absence of key data across many studies. Following consultation with the third reviewer (TF), it was agreed that rather than attempting to impute or exclude studies with missing data, all available data would be presented as reported, and instances of missing data would be transparently noted in the results.

Quality assessment

The methodological quality and risk of bias (RoB) of each included study were systematically evaluated to ensure a robust assessment of internal validity, adhering to PRISMA guidelines for systematic reviews. Study designs were classified through detailed examination of the reported methodology and statistical approaches, following standardised definitions of study designs as outlined by Grimes and Schulz [34]. Studies were categorised as cohort studies only if they employed analytical statistics (e.g., relative risks, hazard ratios, or regression analyses) to compare outcomes between groups, indicating an analytical hypothesis. Studies with two groups that relied solely on descriptive statistics (e.g., means, percentages) to report clinical or laboratory features of HFRS were classified as comparative case series, reflecting their lack of analytical comparison [35].

The aim of this review was to collate clinical and laboratory data on HFRS patients, with all included studies deemed to have sufficient data for this purpose during the screening process. During data extraction, it was determined that calculating measures of effect or establishing causal links was not feasible. Consequently, the review focused on describing clinical and laboratory data as a whole, and the quality of studies did not preclude the extraction of available data from any study.

Quality assessment was conducted using the Joanna Briggs Institute (JBI) suite of critical appraisal tools, selected based on the identified study design [36–40]. For one interventional cohort study with a before-after design and no control group, the National Institutes of Health (NIH) Quality Assessment Tool for Before-After (Pre-Post) Studies With No Control Group was used [41]. Each tool consists of a checklist with questions addressing key bias domains, including selection, confounding, measurement, and statistical validity. Responses to each question were recorded as “Yes,” “No,” “Unclear,” or “Not Applicable" (NA).

To quantify methodological quality, a scoring system was used, tallying the number of “Yes” responses relative to the total applicable items in each checklist. Studies were assigned a quality grade based on this score: Grade 3 (highest quality, > 70% “Yes”), Grade 2 (50–70% “Yes”), and Grade 1 (lowest quality, < 50% “Yes”) [42,43]. It is important to note that the quality grade reflects the study’s adherence to the expected standards for its specific design. For example, a case series with a Grade 3 quality rating meets high standards for case series reporting but remains inherently more prone to bias than a well-conducted cohort study due to differences in study design hierarchy. Thus, a cohort study with a Grade 2 quality rating may still provide more robust evidence than a case series with a Grade 3 rating.

Two independent reviewers (MR and AH) conducted the quality assessments at the outcome level to minimise bias and ensure consistency. Discrepancies were resolved through discussion, with a third reviewer (TF) consulted when consensus could not be reached.

Data synthesis

A formal meta-analysis of laboratory clotting parameters was not performed due to substantial clinical and methodological heterogeneity across the included studies. In addition to inconsistent reporting of central tendency measures and the absence of raw datasets, there were important differences in study populations, infecting hantavirus species, timing of laboratory measurements relative to disease phase, and definitions of haemorrhagic manifestations. Studies included patients infected with different hantavirus strains which are associated with varying clinical severity and haemostatic abnormalities. Furthermore, laboratory parameters were measured at different stages of illness across studies, limiting comparability between reported values. Definitions and reporting of haemorrhagic manifestations were also inconsistent, with some studies reporting minor mucocutaneous bleeding while others reported only clinically significant haemorrhage. These sources of heterogeneity, together with inconsistent reporting of laboratory parameters and lack of variance measures, precluded reliable quantitative pooling of data, including subgroup meta-analyses for commonly reported coagulation parameters.

Transformation of median data to mean data to calculate heterogeneity statistics was attempted but determined to introduce too much potential error into the sample. Instead, weighted averages were calculated using the following formula where M represents median or mean [44]:

In this method, median or mean values from different studies are averaged, but weighted according to the sample size they are derived from. Due to a general lack of reporting on whether data were normally distributed or not, appropriate statistical tests could not be reliably selected to assess for differences in weighted averages. As such, p-values are not reported for these comparisons. Bubble plots were used to visualise the individual median and mean values reported for the various laboratory parameters. Haemorrhagic manifestations were reported as frequencies and percentages. Weighted average laboratory parameters and haemorrhagic manifestation frequencies are reported for all patients with available data as well as stratified by causative hantavirus. Several studies defined cohorts of severe cases in comparison to non-severe cases, including patients who died, those with severe renal failure, or according to severity scoring systems [45–47]. These cohorts of severe cases have been highlighted in the data reporting. Weighted average data for adults and children are considered separately but visualised together in bubble plots. Graphs were created using GraphPad Prism version 10.3.0 for Mac (GraphPad Software, San Diego, California, USA).

Protocol amendments following registration

Following registration of this review on PROSPERO (CRD42024618760), several minor amendments were made to the protocol to improve the clarity and completeness of the final review. The review title was refined from “Characterising the loss of haemostasis in haemorrhagic fever with renal syndrome (HFRS): A systematic review” to “Bleeding in haemorrhagic fever with renal syndrome: A systematic review characterising the loss of haemostasis in hantavirus infections” to better reflect the review’s focus. The database search strategy was amended to replace the Cochrane Library with Scopus, as the review focused on observational, non-interventional study designs for which Cochrane was unlikely to yield relevant results. In addition, an extra search filter restricting results to human studies was applied alongside the English-language limitation to enhance relevance.

The use of blood products, listed as an outcome in the original protocol, was not reported due to insufficient data across included studies. During the screening process, the software Rayyan was used to facilitate abstract screening and manage inclusion/exclusion records, which was not specified in the original protocol but introduced to enhance transparency and reproducibility.

Regarding subgroup analyses, the planned analysis by study design was not undertaken due to limited comparative value between methodologies. Instead, a subgroup comparison of adult versus paediatric patients was added, as this was considered more clinically relevant. These amendments did not alter the overall objectives or analytical framework of the review but were made to strengthen methodological rigour and the interpretability of results.

Literature search and quality assessment

Database searches yielded a total of 1,822 publications. After exclusion of duplicates and studies that did not meet screening criteria, 121 remained for in-depth review. After applying inclusion and exclusion criteria, a final total of 55 publications were included, with a total of 7,950 laboratory-confirmed cases of HFRS. S1 Table summarises the details each of the included studies.

Of the 55 studies included in this systematic review, study designs were classified as follows: 35 cohort studies, 14 case series, 4 case-control studies, 1 diagnostic test accuracy study, and 1 interventional cohort study without a control group. Quality assessment revealed that 31 studies (56.4%) were classified as grade 3 (highest quality), 23 studies (41.8%) as grade 2, and 1 study as grade 1 (lowest quality). Detailed study characteristics, including design and quality grading, are summarised in S1 Table.

The overall methodological quality of the included studies varied across designs. Cohort studies generally demonstrated the highest quality, with most using clearly defined inclusion criteria, standardised diagnostic methods, and reliable measurement of exposures and outcomes. These studies frequently reported demographic and clinical data comprehensively and used appropriate statistical analyses. However, some were limited by incomplete follow-up and potential selection bias where recruitment was not clearly consecutive.

Case series, which formed a substantial proportion of the evidence base, typically provided detailed descriptions of clinical features and laboratory findings, often supported by reliable diagnostic confirmation. Nevertheless, their methodological rigour was reduced by small sample sizes, retrospective data collection, and limited information on recruitment processes, raising concerns about representativeness.

Case-control studies performed well in terms of diagnostic confirmation and clear reporting of clinical data, but most lacked sample size justification and did not always provide sufficient detail on matching procedures or strategies to minimise selection bias.

The single interventional study used validated measurement methods and reported outcomes consistently but was constrained by a small sample size and lacked a control group, limiting interpretation. Similarly, the diagnostic test accuracy study applied appropriate reference standards and reported results transparently, but potential bias arose from limited information on participant selection and blinding.

Across all study designs, common strengths included the widespread use of standardised diagnostic methods and detailed clinical reporting. However, frequent limitations involved small sample sizes, retrospective data collection, incomplete reporting of recruitment strategies, and restricted representativeness, all of which should be considered when interpreting findings.

To better understand how study quality contributed to the evidence base for each outcome, the proportion of patients contributing data from studies of different quality grades was examined. Across the overall cohort of 7,950 laboratory-confirmed HFRS cases, 66.2% of patients were derived from grade 3 studies, 33.3% from grade 2 studies, and 0.5% from a grade 1 study.

For most laboratory parameters, the majority of available data originated from higher-quality studies. Platelet count data were predominantly derived from grade 3 studies (75.4%), as were haemoglobin concentration measurements (78.6%). Similarly, prothrombin time, activated partial thromboplastin time, and fibrinogen measurements were largely contributed by grade 3 studies (80.4%, 76.0%, and 81.8%, respectively). D-dimer measurements showed the highest proportion of data originating from grade 3 studies (93.2%). Liver enzyme measurements were also largely derived from higher-quality studies, with 73.6% of aspartate aminotransferase (AST) data and 72.5% of alanine aminotransferase (ALT) data originating from grade 3 studies.

For clinical outcomes, survival data were predominantly derived from grade 3 studies (69.9%), with a smaller proportion from grade 2 studies (29.6%) and minimal contribution from grade 1 studies (0.5%). Reporting of haemorrhagic manifestations was more evenly distributed between grade 3 studies (47.8%) and grade 2 studies (51.3%), with a small contribution from grade 1 studies (0.9%). Data for paediatric patients were derived almost entirely from higher-quality studies, with 94.1% originating from grade 3 studies and the remainder from grade 2 studies.

Demographic and epidemiological data

Cases were reported from a total of 16 countries, predominantly from China (Table 3). 4,046 cases were caused by HTNV, 2,545 by PUUV, 93 by DOBV and 81 by SEOV. In 1,185 cases the causative hantavirus was not reported despite tests confirming acute hantavirus infection. Amongst adult patients, the weighted average median age of 3,510 patients was 43.2 years, and the weighted average mean age of 4,037 patients was 40.5 years (Table 4). Weighted average median and mean ages were similar when stratifying by hantavirus species amongst adult patients. 5,801 adult patients were male (76.6%) and 1,775 were female (23.4%) (Table 5). The weighted average median age of 105 paediatric cases was 12.1 years, and this data was derived entirely from PUUV-induced HFRS (Table 4). 260 paediatric cases were male (69.5%) and 114 were female (30.5%) (Table 5).

Download:

• PNG
  larger image
• TIFF
  original image

Table 3. Number of publications and HFRS cases reported from each country.

https://doi.org/10.1371/journal.pntd.0014524.t003

Download:

• PNG
  larger image
• TIFF
  original image

Table 4. Weighted average median and mean age of adult and paediatric patients.

https://doi.org/10.1371/journal.pntd.0014524.t004

Download:

• PNG
  larger image
• TIFF
  original image

Table 5. Gender distribution of adult and paediatric patients.

https://doi.org/10.1371/journal.pntd.0014524.t005

The majority of patients included in the reviewed studies were hospitalised. Among HTNV cohorts, 3,896 of 4,046 patients (96.3%) were explicitly reported as hospital inpatients, with the remaining cases derived from studies where hospitalisation could reasonably be inferred from the study design and recruitment process. Similarly, all SEOV patients were derived from hospital-based cohorts, and 77 of 93 DOBV patients (82.8%) were explicitly reported as hospital inpatients, with the remaining cases derived from studies where hospitalisation could reasonably be inferred. Among PUUV patients, 2,101 of 2,545 (82.6%) were explicitly reported as hospitalised inpatients, while 158 patients (6.2%) were reported as outpatients in studies that included both inpatient and outpatient populations. The remaining PUUV cases were derived from studies where hospitalisation was likely but not explicitly stated or where admission status could not be determined.

Laboratory clotting parameters

Adult patients.

Average (median or mean) platelet counts were reported from 46 publications, from a total sample size of 6,562 patients. Average PT data was available for 3,026 patients across 10 studies, APTT for 3,180 patients across 12 studies, and fibrinogen for 2,884 patients across 6 studies. Average D-dimer values were reported from only 5 studies, from 2,153 patients. Data on PT, APTT, fibrinogen, and D-dimer were mainly from HTNV cases. AST data was available for 5,095 patients from 31 studies, and ALT for 5,017 patients from 29 studies.

Detailed data on the weighted average median and mean lab parameters of adult patients are contained in S2 Table, and the values discussed here are for adult patients only. Bubble plots in Figs 2–4 depict the distribution of observed median or mean values from various patient cohorts, including paediatric cohorts. Thrombocytopenia was common, with a weighted average median platelet count of 56.2 x 10⁹/L overall. Thrombocytopenia was more severe in HTNV cases compared to PUUV, with a weighted average median platelet count of 39.9 x 10⁹/L vs 94.5 x 10⁹/L, respectively. Severe cases showed more marked thrombocytopenia compared to non-severe cases (16.3 x 10⁹/L vs 58.7 x 10⁹/L) (Fig 2). The weighted average median PT across all cohorts was 12.1 s. PT prolongation was noted in severe cases (14.9 s) compared to non-severe (11.8 s) (Fig 2). APTT was overall slightly prolonged (weighted average median 38.8 s), with more marked prolongation in severe cases compared to non-severe (weighted average median 45.9 s vs 38.3 s) (Fig 2).

Download:

• PNG
  larger image
• TIFF
  original image

Fig 2. Bubble plots of platelet count and coagulation times by individual study cohort.

Panel A shows median platelet counts and panel B shows mean platelet counts. Panel C shows median prothrombin times (PT) and panel D shows mean prothrombin times. Panel E shows median activated partial thromboplastin times (APTT) and panel F shows mean activated partial thromboplastin times. Each point represents a median or mean value from an individual study cohort. The size of each point represents the sample size on a natural logarithmic scale. Log(sample size) of 7.54 corresponds to a raw sample size of 1,873, and Log(sample size) of 0.69 corresponds to a raw sample size of 2. Bubble colour denotes causative hantavirus. Points denoted with a ‘←’ symbol represent severe cohorts, and points denoted with a ‘⋇’ symbol represent paediatric cohorts. Grey dotted lines illustrate reference ranges and thresholds of clinical importance, with arrows indicating the direction of abnormality. (APTT = activated partial thromboplastin time; DOBV = Dobrava virus; HTNV = Hantaan virus; PT = prothrombin time; PUUV = Puumala virus; SEOV = Seoul virus).

https://doi.org/10.1371/journal.pntd.0014524.g002

Download:

• PNG
  larger image
• TIFF
  original image

Fig 3. Bubble plots of fibrinogen, D-dimer, and haemoglobin concentrations by individual study cohort.

Panel A shows median fibrinogen concentrations and panel B shows mean fibrinogen concentrations. Panel C shows median D-dimer concentrations and panel D shows mean D-dimer concentrations. Panel E shows median haemoglobin concentrations and panel F shows mean haemoglobin concentrations. Each point represents a median or mean value from an individual study cohort. The size of each point represents the sample size on a natural logarithmic scale. Log(sample size) of 7.54 corresponds to a raw sample size of 1,873, and Log(sample size) of 0.69 corresponds to a raw sample size of 2. Bubble colour denotes causative hantavirus. Points denoted with a ‘←’ symbol represent severe cohorts, and points denoted with a ‘⋇’ symbol represent paediatric cohorts. Grey dotted lines illustrate reference ranges and thresholds of clinical importance, with arrows indicating the direction of abnormality. (DOBV = Dobrava virus; HTNV = Hantaan virus; PUUV = Puumala virus; SEOV = Seoul virus).

https://doi.org/10.1371/journal.pntd.0014524.g003

Download:

• PNG
  larger image
• TIFF
  original image

Fig 4. Bubble plots of haematocrit and liver enzyme concentrations by individual study cohort.

Panel A shows median haematocrit levels and panel B shows mean haematocrit levels. Panel C shows median aspartate aminotransferase (AST) concentrations and panel D shows mean aspartate aminotransferase concentrations. Panel E shows median alanine aminotransferase (ALT) concentrations and panel F shows mean alanine aminotransferase concentrations. Each point represents a median or mean value from an individual study cohort. The size of each point represents the sample size on a natural logarithmic scale. Log(sample size) of 7.54 corresponds to a raw sample size of 1,873, and Log(sample size) of 0.69 corresponds to a raw sample size of 2. Bubble colour denotes causative hantavirus. Points denoted with a ‘←’ symbol represent severe cohorts, and points denoted with a ‘⋇’ symbol represent paediatric cohorts. Grey dotted lines illustrate reference ranges and thresholds of clinical importance, with arrows indicating the direction of abnormality. (ALT = alanine aminotransferase; AST = aspartate aminotransferase; DOBV = Dobrava virus; HTNV = Hantaan virus; PUUV = Puumala virus; SEOV = Seoul virus).

https://doi.org/10.1371/journal.pntd.0014524.g004

Fibrinogen levels were generally normal (weighted average median 2.57 g/L), though mild hypofibrinogenemia was noted in severe HTNV cases (weighted average median 1.75 g/L) (Fig 3). Average D-dimer values were consistently elevated in the few studies that reported it, with a weighted average median of 1,810.3 ng/mL (Fig 3).

Average haemoglobin and haematocrit levels were generally within normal limits (Figs 3 and 4 respectively). Liver function showed moderate derangements, with a weighted average median AST of 85.7 U/L and ALT of 56.1 U/L across all patients. Weighted average median AST values were higher in HTNV patients (97.0 U/L) compared to PUUV (41.3 U/L) (Fig 4), while differences in weighted average median ALT were less pronounced (58.7 U/L for HTNV vs 45.4 U/L for PUUV) (Fig 4). Both AST and ALT were higher in cohorts of severe HFRS caused by HTNV (Fig 4).

Haemorrhagic manifestations

Adult patients.

Table 6 contains the frequency and percentages of various haemorrhagic manifestations stratified by causative hantavirus amongst adult patients. It is important to note that the sample sizes this data were derived from were significantly larger for PUUV and HTNV cases compared to SEOV and DOBV. Microscopic haematuria was a relatively common observation across all hantaviruses, present in 28.4% of SEOV cases, 26.7% of DOBV cases, 24.3% of PUUV cases, and 15.8% of HTNV cases. Macroscopic haematuria was reported more frequently in SEOV cases (19.8%), compared to HTNV (2.3%), PUUV (1.5%), and DOBV (0.0%).

Download:

• PNG
  larger image
• TIFF
  original image

Table 6. Haemorrhagic manifestations stratified by causative hantavirus in adult patients.

https://doi.org/10.1371/journal.pntd.0014524.t006

Petechiae of the skin were observed in 24.7% of SEOV cases, compared to 8.5% of PUUV cases and 7.1% of HTNV cases. No petechiae were recorded for DOBV cases. Ecchymoses of the skin were most frequently observed amongst the HTNV (16.7%) and DOBV (13.3%) cohorts and were infrequent in the PUUV cohort (0.2%) and absent in the SEOV cohort. Epistaxis was recorded in all cohorts, most frequently in SEOV cases (9.9%) and infrequently across the other three hantaviruses. Bleeding of the oropharynx was recorded in 12.3% of both the HTNV and SEOV cohorts, which was more common than the DOBV (5.0%) and PUUV (0.4%) cohorts. Ocular bleeding (typically subconjunctival and conjunctival haemorrhages) was recorded in 16.0% of SEOV cases, 10.3% of HTNV cases, 2.7% of PUUV cases, and no DOBV cases.

With regards to more significant haemorrhagic manifestations, gastrointestinal bleeding was the most common site of bleeding. 22.2% of SEOV cases had evidence of gastrointestinal bleeding, compared to 7.4% of HTNV cases, 6.7% of DOBV cases and 2.4% of PUUV cases. Pulmonary and intracranial haemorrhages were rarely recorded across all hantaviruses (Table 6). Additional sites of both minor and major bleeding were recorded either in small numbers or could not be categorised due to ambiguity of reporting. These were recorded as ‘other’ haemorrhagic manifestations, and the specific breakdown of these is shown in S4 Table.

Paediatric patients.

No data on haemorrhagic manifestations were reported for any paediatric cases of HFRS caused by DOBV or SEOV. Data were available for small sample sizes of PUUV and HTNV cases and this is shown in Table 7. Microscopic haematuria was a very common observation, present in 71.4% of PUUV cases 66.7% of HTNV cases. Macroscopic haematuria was also observed in 60.3% of HTNV cases, but none of the PUUV cases. No paediatric cases of PUUV or HTNV had any recorded instances of skin petechiae, ecchymoses or bleeding from puncture sites. Epistaxis and oropharyngeal bleeding were observed amongst 5.7% and 1.9% of PUUV cases, respectively, but no HTNV cases. Ocular bleeding was more frequently observed amongst the HTNV cohort (55.6%) compared to the PUUV cohort (1.0%).

Download:

• PNG
  larger image
• TIFF
  original image

Table 7. Haemorrhagic manifestations stratified by causative hantavirus in paediatric patients.

https://doi.org/10.1371/journal.pntd.0014524.t007

More significant haemorrhagic manifestations were uncommon in paediatric cases of PUUV infection. 2.9% of these cases had gastrointestinal bleeding recorded, and no instances of pulmonary haemorrhage. In contrast, amongst the HTNV cohort, 9.5% had evidence of gastrointestinal bleeding and 7.9% pulmonary bleeding. No paediatric cases of PUUV or HTNV had intracranial bleeding. Four patients within the PUUV cohort were recorded as having ‘other’ haemorrhagic manifestations. These related to two cases recorded as having ‘visible bleeding’ without any site specified, and two cases of abnormal vaginal bleeding.

Clinical outcome

Studies that reported on survival and death were used to calculate case fatality rates (CFR) for each hantavirus (Table 8). The case fatality rate for adult cases of HFRS caused by PUUV, DOBV and HTNV were 0.1%, 10.6% and 6.0% respectively. No deaths were recorded amongst SEOV patients. For paediatric cases, the case fatality rate of HTNV cases was 2.6%. No paediatric cases of PUUV died. No data were reported for paediatric cases of DOBV or SEOV.

Download:

• PNG
  larger image
• TIFF
  original image

Table 8. Case fatality rate of different hantaviruses amongst adult and paediatric patients.

https://doi.org/10.1371/journal.pntd.0014524.t008