Two Cases of Lemierre Syndrome: Does One Definition Fit All?
Case 1
A previously healthy 23-year-old male was admitted for sore throat, neck pain, diarrhoea
and vomiting starting 1 week earlier. Physical examination revealed left tonsil enlargement,
neck swelling and chronic periodontitis. In the suspicion of bacterial sepsis (fever,
tachycardia, high respiratory rate, leukocytosis [16.1*109/L], elevated inflammatory biomarkers [C-reactive protein 32.0 mg/L; procalcitonin
276.7 µg/L]), the patient was started on piperacillin–tazobactam, levofloxacin, steroids
and fluid resuscitation. Despite treatment, he developed severe hypotension, hypoxaemia,
acute renal failure and thrombocytopenia. Neck compression ultrasound showed left
internal jugular vein thrombosis; computed tomographic pulmonary angiography (CTPA)
depicted left superior thyroid vein thrombosis and bilateral pulmonary abscesses ([Figs. 1] and [2]). Antibiotic therapy was switched to meropenem plus metronidazole, and prophylactic-dose
enoxaparin. A prophylactic dose of low molecular weight heparin (LMWH) was preferred
in the presence of conditions reducing its clearance (renal failure) and increasing
the individual risk of bleeding (thrombocytopenia). An anaerobic gram-negative rod
of the Prevotella spp. was isolated from blood cultures. The patient's clinical condition progressively
improved. Three weeks later, CTPA showed reduction in the size of the pulmonary nodules
and resolution of jugular and thyroid vein thromboses. The patient was discharged
on amoxicillin–clavulanate, ciprofloxacin and prophylactic-dose enoxaparin, which
was discontinued after 6 weeks from discharge after completion of approximately 3
months of treatment, without further complications.
Fig. 1 Clinical case 1: Thrombosis of the left superior thyroid vein (arrow) associated
with lymphadenomegaly of the left neck (arrowhead).
Fig. 2 Clinical case 1: Computer tomography pulmonary angiogram sections showing multiple
bilateral nodules within the lung parenchyma consistent with infarctions evolving
to abscesses (both panels).
Case 2
A 40-year-old male presented with a 5-day history of fever and neck swelling associated
with leukocytosis (20.1*109/L) and elevated inflammatory biomarkers (C-reactive protein 20.4 mg/L; procalcitonin
14.2 µg/L). He reported chronic hepatitis B virus infection, hypertension, diabetes
mellitus and recent tooth extraction. CTPA detected right internal jugular vein thrombosis,
lymphadenomegaly, right laterocervical abscess and focal ground glass alterations
in the ventral segment of the upper lobe of the right lung ([Fig. 3]). Incision and drainage of the abscess was performed and the Gram-negative Klebsiella pneumoniae was isolated from cultures of the drainage material. The patient received meropenem,
vancomycin and therapeutic-dose nadroparin. CTPA performed 6 weeks after admission
demonstrated complete recovery of the previous alterations. The patient was discharged
on trimethoprim-sulfamethoxazole and therapeutic-dose nadroparin. Anticoagulation
was given for a total of 3 months without complications.
Fig. 3 Clinical case 2: Internal jugular vein thrombosis and large right laterocervical
abscess. (A) Thrombosis of the right internal jugular vein (arrow). Contrast enhancement and
thickening of the vessel wall indicates venous thrombosis associated with signs of
inflammation and oedema of the perivascular adipose tissue (arrowhead). (B) Thrombosis of the right internal jugular vein (arrow) and large right laterocervical
abscess (arrowhead).
Lemierre Syndrome: An Update
How to Define and Diagnose Lemierre Syndrome
André Lemierre was among the first physicians to provide a comprehensive description
of the clinical features of what he called ‘anaerobic post-anginal septicaemia’ in
1936,[1] named after him (but only 50 years later) ‘Lemierre syndrome’,[2] or alternatively, with the synthetic genitive, ‘Lemierre's syndrome’. He described
the syndrome as ‘usually affecting young adults or adolescents, with both sexes being
equally attacked’, sometimes occurring in small epidemics, with ‘the most usual initial
cause being a tonsillar or peritonsillar abscess, opened too late or to an insufficient
degree’. Moreover, ‘a simple tonsillitis may conceal small foci of suppuration. […]
These septicaemias are the result of a thrombophlebitis of the tonsillar and peritonsillar
veins, which can spread to the internal jugular vein or even to the facial vein. […].
The B. funduliformis septicaemias have always been accompanied by the formation of distant metastatic
abscesses’, as well as by multiorgan failure, and can rapidly progress to death.[1] Despite this early and clear recognition of a new nosological entity, there is still
no agreement regarding the criteria required for confirming the diagnosis of Lemierre
syndrome.[3]
[4] This may partly explain the delayed introduction as a separate Medical Subject Heading
(Unique MeSH) ID D057831 in 2011 and the persisting absence of a specific International
Classification of Diseases code for its identification.
Several definitions, including the initial one made by Lemierre himself, pinpoint
the varying characteristics of the vascular involvement (venous thrombosis and/or
septic embolism), organ failure and anaerobic bacterial species detected from local
source of infection or from an otherwise sterile site. Stricter criteria are suggested
in the narrative review authored by Riordan in 2007,[5] which focused on a total of 222 cases, namely: (1) history of [tonsillar] anginal
illness or compatible clinical findings; (2) evidence of metastatic lesions in lungs
and/or another site; and (3) either evidence of internal jugular vein thrombophlebitis
or isolation of F. necrophorum, a strict anaerobe Gram-negative rod, or Fusobacterium spp., from blood cultures or from a normally sterile site. As previously mentioned, more
liberal definitions or variants are found in the literature, which include other head
and neck infectious foci than tonsils and widely heterogeneous thromboembolic complications,
ranging from cerebral sinus vein to carotid artery involvement, presenting alone or
in association with internal jugular vein thrombosis.[6] Furthermore, as highlighted by our cases and in previous reports, other anaerobic
or aerobic bacteria, such as Prevotella spp. (Case 1)[7]
[8] or Klebsiella pneumonia (Case 2),[9]
[10]
[11]
[12] are associated with similar clinical pictures. Whether these are truly causative
agents, or F. necrophorum undetected, remains unknown.[13]
[14] Importantly, differences in the definition of the diagnostic criteria of Lemierre
syndrome do not ultimately translate to different diagnostic and therapeutic approaches.
The clinical suspicion of Lemierre syndrome should be raised—ideally before the signs
of sepsis become manifest—in teenagers or young adults presenting with sore throat,
high temperature, chills and unilateral tenderness of the neck. This is especially
the case if the rapid streptococcal antigen from a tonsillar swab is negative and
in the presence of signs of multifocal pneumonia on a chest X-ray. A computed tomography
(CT) scan of the neck and chest may facilitate the diagnosis of neck vein thrombosis
and depict typical rounded multifocal infiltrates suggestive for septic pulmonary
emboli.
An algorithm consisting of the sequential application of a clinical decision score,
D-dimer testing and ultrasonography has been validated for the diagnosis of upper-extremity
deep venous thrombosis, including jugular vein thrombosis; however, patients with
abscesses and massive lymphadenomegaly, such as those with Lemierre syndrome, were
under-represented in the derivation study.[15] The use of positron emission tomography with 2-[fluorine-18]fluoro-D-glucose integrated
with CT (FDG-PET/CT) has been described in three patients with Lemierre syndrome and
may represent an alternative or adjunct to conventional imaging in selected patients.[16] Blood cultures usually confirm the diagnosis, being positive for anaerobic bacteria
in most cases, together with a tonsillar swab cultured in selective anaerobic plates.
Specific molecular detection using polymerase chain reaction techniques can add useful
information.[17]
Incidence and Case Fatality Rates
Only a few studies have investigated the epidemiology of Lemierre syndrome, which
remains an elusive disease because of its rarity. All the same, Lemierre syndrome
is not ‘a forgotten disease’, as previously thought and often referred to.[18] Clinical experience suggests that the incidence of Lemierre syndrome may be on the
rise, and it remains to be seen whether this is due to antibiotic resistance, changes
in antibiotic prescription patterns, more efficient diagnostics and/or publication
trends ([Fig. 4]).
Fig. 4 Publication trends illustrating the annual number of studies reported in PubMed for
Lemierre syndrome and Fusobacterium spp. between years 1988 and 2017.
The most important prospective epidemiological study was conducted in Denmark between
1998 and 2001; it estimated an annualized incidence rate of 3.6 cases per million
in the general population.[19] In this nationwide study, the authors included all disseminated F. necrophorum infection cases with primary foci in the head and neck: the highest incidence was
found in individuals aged 15 to 24 years old with 14.4 cases per million/year, while
the disorder was rare in adults ≥ 40 years old. Previous reports suggested lower annualized
incidence of approximately one event per million general population.[20]
[21]
A second Danish prospective analysis was conducted between 2010 and 2014, and focused
on the incidence of F. necrophorum bacteraemia from microbiology samples.[22] Consistently with what was previously reported, the estimated annualized incidence
was higher in young individuals (9.4 cases per million patient-years) compared with
children and middle-aged patients. With an overall estimated annualised incidence
of 13.9 cases per million patient-years, they suggested an unchanged incidence in
Denmark over the past two decades.[22] Finally, some narrative reviews suggested higher probability of the syndrome in
males with a 2:1 ratio,[5]
[23]
[24] which was, however, not confirmed by others,[25] with seasonal peaks in early autumn and late winter.[3]
The overall case fatality rate was 9% with peaks of 26% in the elderly.[19] The elevated risk of death in the elderly was subsequently confirmed.[22] A narrative review of the literature attempting to focus on deaths unrelated to
underlying disease, such as immunodeficiency or cancers, estimated an overall case
fatality rate of 4 to 12%,[26] in line with what was observed (5%) in a systematic review of 84 cases.[3] Indeed, one should be cautious in interpreting results based on a post hoc categorization
of the causes of death.
Causes of Acute Vein Thrombosis in Lemierre Syndrome
Adult patients with acute bacterial infections have an approximately threefold increased
risk of developing venous thromboembolism compared with patients without infection,
especially within the first 2 weeks after onset.[27]
[28] However, the absolute rate of thromboembolic complications remains low overall and
no other infectious condition apart from the Lemierre syndrome presents with thromboembolic
complications as an integral part of the initial clinical picture and diagnostic criteria.
The jugular vein is commonly affected, likely due to its location in the parapharyngeal
space adjacent to the tonsils.[3]
[5] The thrombosis may be found also in the distal part of the jugular vein, and sometimes
in the external jugular vein.[6] It is therefore likely that the infection spreads not only via the tonsillar vein
(into the lingual and pharyngeal veins, and into the internal jugular vein), but also
through the tissues or via the lymphatic system leading to inflammation of the vessel
wall, acute thrombosis and bacterial invasion.
F. necrophorum, the most common pathogen in Lemierre syndrome, has been shown to assemble and activate
the pro-coagulant and pro-inflammatory contact system on its surface, leading to activation
of the intrinsic pathway of coagulation and release of bradykinin, which in turn may
favour thrombus stabilization and cause increased vascular leakage.[29] F. necrophorum also binds to plasminogen, which is more readily activated to plasmin at the bacterial
surface than unbound plasminogen; moreover, the plasmin bound to the surface is protected
from inactivation by α2-antiplasmin.[30] This can lead to increased invasion through extracellular tissue and adherence to
endothelial cells.[31] Indeed, many other pathogenic bacteria that are able to directly promote thrombosis,
such as Staphylococcus aureus,[32] can activate plasminogen, breach tissue barrier and more effectively spread and
evade innate immune defense,[33] supporting the concept of Lemierre syndrome not being exclusively caused by F. necrophorum.
In addition to Lemierre syndrome, the presence of F. necrophorum is also associated with localized pharyngotonsillitis and peritonsillar abscesses,
preferably occurring in the same young age group.[34]
[35] The causes of age distribution of Lemierre syndrome could involve a recent acquisition
of the bacterium combined with individual lack of adaptive immunity, much in the same
way as meningococcal disease affects non-immune teenagers worldwide.[36] In addition to the adaptive immunity, modulation of innate immunity by the bacteria
may be important for the invasion to occur. Most F. necrophorum strains are able to bind factor H to their surfaces and in a plasma environment this
may contribute to the protection against complement attack by preventing (or limiting)
complement activation.[37] Moreover, F. necrophorum isolated from animals, and also found in human isolates,[38]
[39] has long been known to secrete a leukotoxin that kills neutrophils and is considered
a major virulence factor in animal disease due to F. necrophorum.[40] Theoretically, it may impair local innate immunity and increase the invasiveness
of the bacterium.
A recent Swedish study found that in two-thirds of the F. necrophorum positive pharyngotonsillitis patients, another bacterium or virus was also detected.[41] In a cohort of 22 patients with Lemierre syndrome, the prevalence of serologically
confirmed active Epstein–Barr virus infection was not higher (4.5%) than one would
expect in a paediatric population with pharingotonsillitis.[6]
In addition to pro-coagulant properties of the bacteria, host factors might also play
a role in the genesis of thrombosis. The screening of 26 patients with Lemierre syndrome
provided no conclusive data regarding the possible role of underlying thrombophilia.[6]
How the Clinical Presentation of Lemierre Syndrome Influences Prognosis
As discussed, the most frequent characteristics of Lemierre syndrome are a history
of recent oropharyngeal infection, clinical or radiological evidence of neck vein
thrombosis and isolation of usually anaerobic pathogens. However, there is often a
fourth feature, the presence of distant metastatic emboli, associated or not with
severe sepsis and septic shock, that raises the alarm and points towards the diagnosis.[3] Neck vein thrombosis can lead to septic embolization via haematogenous spread. Lung
involvement, followed by septic arthritis/osteomyelitis, deep neck space infections
and liver, spleen, kidney and brain embolization are the most common manifestations
affecting distant organs.[3]
[5] If not diagnosed and ‘interrupted’ early, this phenomenon of distant bacterial spread
increases morbidity and mortality and directly affects prognosis propagating the thrombo-inflammatory
process leading to multiorgan failure.
Blood cultures can often be negative or take as long as 7 days for identification
of F. necrophorum in less advanced bacteriological laboratories,[42] and the same goes for tonsillar cultures. Imaging tests showing neck vein thrombosis
or signs of septic embolism often provide the initial and only clue for suspecting
Lemierre syndrome. A history of oropharyngeal infection, especially in previously
healthy young adults or children, not responding to ‘conventional’ antibiotics within
24 hours of intravenous treatment, should alert the clinician for a multidisciplinary
team approach. Early surgical treatment, when indicated, and aggressive intravenous
antibiotics targeting anaerobes appear reasonable therapeutic options if any of the
diagnostic features of Lemierre syndrome are present. As for severe sepsis and septic
shock,[43] the prognosis of Lemierre syndrome may largely depend on timely initial management,
which should be ideally completed within 6 hours after the patient's presentation
to minimize adverse outcomes.[44]
[45] In a recent study of more than 49,000 patients with sepsis and septic shock, a rapid
completion of a 3-hour course of sepsis care and administration of antibiotics, but
not rapid completion of an initial bolus of intravenous fluids, were associated with
lower mortality.[46]
Controversy over the Benefits versus Risks of Anticoagulants
The efficacy and safety of anticoagulant drugs for the treatment of vein thrombosis
associated with Lemierre syndrome are uncertain.[47] While prompt institution of anticoagulation may prevent thrombotic extension and
embolization and new thromboembolic events, this treatment may expose patients to
high risk of bleeding complications in the presence of concomitant bleeding risk factors
(e.g. thrombocytopenia or worsening renal function due to sepsis). Moreover, it remains
unclear whether anticoagulation may facilitate the spread of septic material and/or
haemorrhagic transformation of embolic lesions, as it may be the case for the initiation
of de novo anticoagulation in patients with infectious endocarditis.[48] The limited evidence on the pharmacokinetics of anticoagulants in childhood or adolescence
represents an additional challenge, although new therapeutic options and regimens
are being studied, especially for long-term or extended treatment.[49]
[50]
[51]
[52]
When deciding on the agent and optimal dosing, physicians should carefully consider
drug pharmacokinetics, individual patient characteristic, disease severity, bleeding
risk and individual risk of new thromboembolic complications. For instance, it appears
reasonable to prefer LMWH over fondaparinux in patients with moderate renal impairment
because of their shorter half-life and the lower renal clearance. In the acute phase,
unfractionated heparin may be considered in case of severe renal insufficiency and
in the management of sepsis, since progression to severe sepsis and septic shock may
be swift. The presence of mucosal lesions or inflammation, nausea and vomiting may
decrease oral anticoagulant intake while concomitant or ensuing gastrointestinal disturbances
(e.g. diarrhoea) may affect drug absorption.
Patients with clinical improvement may be transitioned to oral anticoagulation, which
can be continued until up to 3 months as recommended by major clinical guidelines
for the cases of provoked venous thrombosis.[53] Longer courses of oral anticoagulation with either vitamin K antagonists or novel
direct oral anticoagulants should be considered for patients who respond poorly (i.e.
in terms of thrombus resolution), remain symptomatic despite antibiotic treatment
or in whom the infectious process is not completely resolved. A periodic re-assessment
of the risks of bleeding and recurrent thrombosis is warranted to decide on the optimal
treatment duration,[53] although no validated risk assessment models are available for the paediatric population
and several of the predictors identified in adults, such as arterial hypertension,
cancer and chronic renal dysfunction, would not apply.
Designing and Conducting a Systematic Review and Individual Patient Data Meta-Analysis
Review Questions
-
What is the rate of thromboembolic recurrence, thrombotic extension of index thrombosis
or new embolism in patients diagnosed with Lemierre syndrome?
-
What are the major bleeding and fatality rates in patients diagnosed with Lemierre
syndrome?
-
What are the estimated efficacy and safety of anticoagulant treatment?
Eligibility Criteria
The following diagnostic criteria for Lemierre syndrome served for the selection of
potentially eligible studies: (1) the primary site of infection being located at either
the head or the neck sites; (2) isolation of F. necrophorum or other bacterial organism potentially causing Lemierre syndrome, associated or
not with the presence of sepsis; and (3) neck/cerebral/upper-extremity thrombosis
and/or septic embolism ([Table 1]). This definition of Lemierre syndrome aims to focus on a broad spectrum of cases
originated from head/neck bacterial infections and associated with characteristics
thromboembolic complications to perform predefined analyses in relevant patient sub-groups.
Such definition does not cover cases originated from sub-diaphragmatic infections
(i.e. urogenital or gastrointestinal infections, or liver abscesses) since septic
acute thrombosis usually affects older individuals and might be due to different etiopathology,
including impaired mucosal integrity due to cancer.[54]
Table 1
Summary of the study protocol of the systematic review and individual patient data
meta-analysis
|
Research questions
|
1. What is the rate of thromboembolic recurrence, thrombotic extension of index thrombosis,
or new embolism in patients diagnosed with Lemierre syndrome?
|
|
2. What is the rate of major bleeding and fatality rate in patients diagnosed with
Lemierre syndrome?
|
|
3. What are the estimated efficacy and safety of anticoagulant treatment?
|
|
Diagnostic criteria accepted for Lemierre syndrome
|
• The primary site of infection being located at either the head or the neck sites
|
|
• Isolation of F. necrophorum or other bacterial organism potentially causing Lemierre syndrome, associated or
not with the presence of sepsis
|
|
• Neck/cerebral/upper-extremity thrombosis and/or septic embolism
|
|
Study outcomes
[a]
|
• Recurrent venous thromboembolism (new thrombotic events or extension/recurrence),
and new septic embolism
|
|
• Major and non-major bleeding (ISTH criteria)[55]
|
|
• Death from all causes
|
|
Subgroup analyses
|
• Paediatric versus adult population
|
|
• Patients grouped by use and type of anticoagulant agent
|
|
• Patients undergoing surgery versus not
|
|
• Type of antibiotic treatment received and type of bacteria involved
|
|
• By geographical region
|
|
• Based on the results of quality assessment
|
|
Registration of the study protocol
|
International prospective register of systematic reviews PROSPERO (ID CRD42016052572)
|
Abbreviation: ISTH, International Society on Thrombosis and Haemostasis.
a Assessment of all outcomes is considered either during hospitalization or at a short-term
follow-up (30 days).
Literature search covered the period elapsing from January 2000 (week 1) to October
2016 (week 3). Only studies published after year 1999 were included since that date
approximately represents the time since LMWH became commonly prescribed worldwide
for parenteral anticoagulation. We applied neither language nor study design restrictions
to the literature search, as we anticipated that the vast majority of studies consist
of case reports or small case series.
Literature Search Strategy and Data Extraction
We systematically searched MEDLINE (via PubMed), Embase (via Ovid) and the Cochrane
Library database on 19 October 2016 ([Table 2]). Our literature search identified a total of 1,971 studies (n = 1,577 for the timeframe selected and n = 1,012 after removing duplicates). Details of our literature search strategy are
provided in the [Supplementary Material]. We complemented this search by manually reviewing references of retrieved articles,
review papers, guidelines and grey literature (Google Scholar, ResearchGate). In case
of duplicates or multiple congress abstracts, we decided to use the most recent full-text
paper.
Table 2
Phases of the systematic review and patient level analysis
|
Study design
|
Definition of the research questions, study population of interest and study outcomes
(study protocol and registration)
|
|
Development of the electronic database for data extraction
|
|
Development of the literature search strategy
|
|
Literature search in PubMed, Embase and the Cochrane Collaboration repository
|
|
Selection of studies
|
Evaluation for eligibility based on titles and abstracts
|
|
Retrieval of full texts of the selected studies
|
|
Second selection based on full texts
|
|
Retrieval of additional studies from grey literature and cross-referencing
|
|
Data extraction
|
|
Quality assessment
|
Quality assessment of the included studies
|
|
E-mail contact with all the authors of the studies to provide an estimate of publication
bias and to retrieve missing data
|
|
Data analysis
|
Data cleaning
|
|
Adjudication of the study outcomes
|
|
Statistical analysis
|
Titles and abstracts were screened by three reviewers, who selected titles subsequently
assessed for eligibility after evaluation of full-texts. Disagreements were solved
by consensus. Neither of the reviewers was blind to the journal titles or to the study
authors or institutions. Corresponding authors were contacted by email within at least
2 weeks apart to provide missing data or to clear ambiguity about original data, if
crucial information were lacking and to retrieve additional information regarding
the outcomes of interest.
The following variables were collected: age, sex, year, type of article and number
of patients included, cancer (presence and type), infection of head and/or neck (presence
and type), isolation of F. necrophorum, isolation of other pathogens (type or phyla), utilization and type of antibiotics, evidence of jugular vein thrombosis at the
time of Lemierre syndrome diagnosis, thrombosis in other head/neck districts, description
and treatment of first event of thrombosis (yes/no, kind of anticoagulant treatment,
dosage, days of treatment), clinical outcomes (occurrence, type of anticoagulant treatment,
time to event), surgical treatment, delay in diagnosis and length of hospitalization.
Study Outcomes
Recurrent thromboembolic events were defined as new thrombotic events or extension/recurrent
thrombosis occurring at the same site ([Table 1]). The occurrence of new septic embolism was defined on the basis of the description
provided in the original paper. Bleeding events were retrospectively classified as
‘major’ or ‘non-major’ (International Society on Thrombosis and Haemostasis criteria).[55] Major bleeding includes fatal bleeding, and/or symptomatic bleeding in a critical
area or organ, such as intra-cranial, intra-spinal, intra-ocular, retroperitoneal,
intra-articular or pericardial or intra-muscular with compartment syndrome, and/or
bleeding causing a fall in haemoglobin level of 20 g L−1 or more, or leading to transfusion of two or more units of whole blood or red cells.
Reasons of death are categorized as Lemierre syndrome-related (sepsis, thromboembolic/bleeding
complications, post-surgical complications) or not. Assessment of all outcomes was
considered either during hospitalization or at a short-term follow-up (30 days). Information
on all outcomes was collected in a separate case report form prepared by a third investigator
and adjudicated independently based on the aforementioned criteria by two investigators
unaware of the anticoagulant treatment status described in the original studies.
Quality Assessment and Risk of Bias
The included studies were primarily categorized according with their study design
(case reports, small series of cases, cohort studies), the presence of consecutive
patients and the proportion of missing data. To minimize the risk of publication bias,
all the authors of the included studies are contacted and asked whether some unspecified
selection criteria were applied to the published report and if they have treated other
patients with Lemierre syndrome before, after or concomitantly to the current publication
([Table 2]). As previously mentioned, relevant papers in the field and grey literature were
screened to detect additional publications, since traditional databases may not include
all case reports, particularly those not undergoing a peer-review process.[56]
Registration of the Study Protocol and Status
The study is registered in the international prospective register of systematic reviews
PROSPERO (available at crd.york.ac.uk/PROSPERO) on 1 December 2016 and updated on
10 August 2017 for status of the review and minor corrections (ID CRD42016052572;
[Table 2]). After completion of data retrieval and quality assessment, we expect to complete
the primary analysis by June 2018.
Statistical Analysis
Aggregate data will be analysed at the level of individual patients and a quantitative
synthesis is planned. Rates of events occurring within 30 days (or in-hospital) will
be provided as proportions and 95% confidence intervals. Predictors of 30-day adverse
outcomes will be evaluated in exploratory analyses by logistic regression models.
Sensitivity analyses accounting for studies grouped according with study design, quality
assessment and proportion of missing data are planned. Predefined sub-group analyses
will include: (1) paediatric versus adult population; (2) patients grouped by type
of anticoagulant agent; (3) patients undergoing surgery versus not; (4) type of antibiotic
treatment received and type of bacteria involved; (5) by geographical region; and
(6) based on the results of quality assessment ([Table 1]). Multiple imputation techniques will be considered for overcoming the problem of
cases with missing values.
The general methodology for performing case reports (or cases series)-based analyses
has been adapted from available studies with similar design and aims investigating
rare conditions.[57]
[58]
[59]
[60]
[61]
[62] The recommendations listed in the Preferred Reporting Items for Systematic Review
and Meta-Analyses of individual patient data statement will be followed, where applicable,
for reporting the results of this analysis.[63]
