Study design and patients

We conducted a retrospective multicenter cohort study in the Paris–Ile-de-France region, which represents a population of nearly 10 million inhabitants, by studying the databases of nine French university hospitals. The participant centers were the Bicêtre, Lariboisière, Saint-Joseph, Mignot, Bichat, Sud Francilien, Rothschild, Foch, and Sainte-Anne hospitals. We included every patient with a first-ever diagnosis of symptomatic CVT by magnetic resonance imaging (MRI) or computed tomography (CT) scan, who were aged ≥65 years and had been hospitalized in a stroke unit or a neurological department between January 2011 and November 2021. Patients were screened via a search of the diagnosis according to the International Classification of Diseases, 10th Revision (ICD-10; thrombophlebitis, cerebral infarction due to CVT) in the participating hospitals (Programme de médicalisation des systèmes d'information [PMSI], the French national hospital discharge database), providing a comprehensive list of target patients. CVT definition comprised evidence of thrombosis on venous sequences (time of flight and T1 gadolinium) on MRI or on CT venography.

Standard protocol approvals, registrations, patient consent, and ethics

This study was approved by the local institutional review board/independent ethics committee (IRB/IEC; CPP-VII). As this is a retrospective observational study with anonymized data, without any additional therapy or monitoring procedure, the IRB/IEC did not identify ethical issues and declared that, in accordance with French legislation, formal approval was not required.

Data collection

We reviewed medical records of all identified patients using a standardized approach to collect several data types: demographic data, preexisting medical conditions, clinical presentation, time between symptom onset and diagnosis, treatment modalities, type and results of the diagnostic and etiological workup (including cerebral imaging to determine involved sinus and parenchymal lesions, usual blood tests, cerebrospinal fluid [CSF] analysis, thrombophilia investigation, and any more specialized analyses), retained associated conditions/risk factors, and clinical outcome (at 3 months and 1 year).

Clinical features in the acute phase

We collected the proportion of patients with the following modes of presentation of CVT: confusion, headache, isolated headache, isolated intracranial hypertension (headache with CSF pressure ≥ 25 cm H₂O, or papilledema), focal deficit, seizures, and coma. Symptom onset was defined as acute (<48 h), subacute (<1 month), or chronic (>1 month). Time before medical visit, number of visits before diagnosis, time before diagnosis (defined as time between symptom onset and diagnosis confirmed by imaging), and diagnosis delay (time between first medical visit and diagnosis) were noted when available.

Etiological investigations

We gathered all information regarding associated medical conditions that could be potential underlying risk factors (noting that causality cannot be proven). The medical conditions that were considered relevant were the generally admitted risk factors for CVT, such as active cancer, hemopathy (acute or chronic), JAK2 mutations, nonacquired thrombophilias (antithrombin III, protein S or protein C deficiency, factor V and II mutation), antiphospholipid syndrome (APS; confirmed with 3-month hindsight), local or systemic infection, inflammatory diseases, and/or hormonal exposure.

Outcome

We collected the following information: in-hospital mortality, hemorrhagic complications, and deaths occurring during the follow-up, ability to return home after hospitalization, frequency of motor and cognitive impairment, and clinical outcome at 3 months and 1 year after CVT, which was assessed by the modified Rankin Scale (mRS). A good functional status was defined as an mRS of 0–1.

Incidence

In France, the national census regularly provides detailed population data according to age/region. We used these data to estimate the population of patients aged ≥65 years in the Paris–Ile-de-France area covered by participating centers during the study period. Annual incidence of CVT in people aged ≥65 years in the aforementioned area was then estimated from the number of patients included in our study compared with the population data.

Data validation and comparison with national data from the hospital discharge database

Considering the retrospective nature of our study and that some hospitals of the Ile-de-France region did not participate, we decided to confirm the reliable identification of all potential cases despite the retrospective nature of our study. Thus, we compared the number and age distribution of first-ever CVT observed in our study to those in the total Ile-de-France and overall French population for the same age category over the corresponding period of time (2011–2021) using data extracted from the French national discharge database (ICD-10 coding) [8]. In France, the national health insurance information system contains individual, anonymized data concerning the beneficiaries of the various national health insurance schemes for the entire French population [8]. Using these databases, we were able to compare the age-specific distribution in our study population with that of the total French and Ile-de-France region population.

Subgroup comparisons

Despite modest sample size, we made exploratory analysis and compared several characteristics according to sex (female/male) or age (dividing our cohort into two groups of people aged ≤75 years or >75 years).

Comparison with the prospective Lariboisière register of adult patients with CVT younger than 65 years

Some characteristics of our cohort (≥65 years old with CVT) were compared with the prospective Lariboisière CVT register of 478 adults aged <65 years, recruited from 1998 to 2020. The Lariboisière register gathered clinical, etiological, management, and prognostic data [2].

Statistics

Statistical comparisons were made for different features. First, we made comparisons among patients aged ≥65 years between male and female patients and comparisons between the youngest and oldest patients (65–74 years vs. ≥75 years). Second, we also compared patients aged ≥65 years and adult patients aged <65 years from the Lariboisière register. Descriptive data are reported associated with the percentage and either mean and SD or median and interquartile range (IQR). Distributions were compared by use of a χ² test when the number of patients per both groups was ≥5 and with the Fisher exact test when the number of patients in at least one group was <5. The Mann–Whitney test was used to analyze mean age among patients aged ≥65 years owing to the nonparametric distribution of ages. We performed a univariate analysis with a test of logistic regression to identify patient characteristics associated with a good prognosis (mRS = 0–1). Results are given with odds ratios (ORs) and 95% confidence intervals (CIs).

Population characteristics and demographic data

One hundred fourteen symptomatic patients aged ≥65 years with a first CVT presenting between January 2011 and November 2021 were identified at the participating hospitals (4–29 patients per center). Most patients were female (61%). The median age at CVT occurrence was 72.6 years (IQR = 67.8–79.0), and the oldest patient was 93 year old (Table 1). The incidence of first-ever symptomatic CVT in patients was 5.9 per million inhabitants aged ≥65 years per year in our cohort according to our analysis of medical records. According to data from the French national discharge database, yearly incidence of first-ever symptomatic CVT was 7.1 per million inhabitants aged ≥65 years in the Ile-de-France population and 8.5 per million in France overall. Figure 1 shows the incidences stratified by age group. In our cohort, as in the Ile-de-France national health insurance database, CVT incidence decreased with age.

Presentation and diagnosis

Characteristics of patients aged ≥65 years are presented in Table 1. Symptom onset was predominantly acute or subacute (46% and 42%, respectively) and rarely chronic (12%). The median time from symptom onset to diagnosis was 5 days, and diagnosis was made in the first 48 h in most cases (64%). The two most common clinical features were headaches (57/114, 50%) and focal deficits (58/113, 51%). More than half of patients experiencing headaches had another associated clinical feature (32/57, 56%). Seizures and confusion were also common (46/114, 40% and 31/114, 27%, respectively). Among patients who presented with seizures, it was the sole manifestation for 30% of them. Funduscopic examination was performed for only 40 patients (35%), and papilledema was identified in seven (18%).

The thrombus location was a lateral sinus in 69% of cases, the superior sagittal sinus in 43% of cases, and in multiple locations in 46% of cases. Cortical vein thrombosis was observed in 32 patients (28% of cases), including six patients without involvement of either dural sinuses or the deep venous system. Fifty-two patients of our cohort (46%) had parenchymal damage, with venous infarction (25%), cerebral hemorrhage (30%), or both venous infarction and cerebral hemorrhage (9%).

Associated risk factors

Hemopathies and cancers were the most common CVT-associated conditions. In cases where there was no history of known cancer, it was searched for by systematic screening by CT or positron emission tomography scan in 73 of 106 patients (69%). Overall, there were 15 cases (14%) of active cancer and eight cases (7%) of hemopathy (acute and chronic). A JAK2 mutation was discovered four of the 18 times it was searched for (22%), plus one patient for whom the mutation was already known, as he had Vaquez polyglobulia. Only three of the five patients with a JAK2 mutation had thrombocytosis or polyglobulia. There were four cases of inflammatory rheumatological disease, three cases of rheumatoid polyarthritis, and one case of spondylarthritis. Lumbar puncture was performed for 35 patients in our cohort; there were only two cases of bacterial meningitis. The ear, nose, and throat (ENT) area infections (8/114, 7%) were mainly sinusitis (5/8, 63%); there were two cases of mastoiditis and one parotiditis with face infection. There were no cases associated with SARS-Cov-2 infections or COVID-19 vaccines. Congenital thrombophilia and antiphospholipid syndrome were searched for in most patients (63%). We discovered nine patients in our cohort who had congenital thrombophilia; there were eight heterozygous mutations of the prothrombin gene and one heterozygous factor V gene mutation, two patients had protein C deficiency, and one had antithrombin deficiency. A diagnosis of APS was made in five cases. A definite associated condition (any of the conditions mentioned above) was found in 45 patients (39%). Only one female patient was receiving hormone replacement therapy (HRT) before CVT diagnosis.

Therapeutic management, outcome, and prognosis

Anticoagulant therapy was administered to 93% of patients. One patient was treated with thrombolysis, thrombectomy, and craniectomy, and another patient was treated with thrombectomy alone. Nine patients of the cohort (8%) had a fatal outcome; all the deaths occurred in the acute phase or during the first 3 months following the CVT. Most patients were able to return home at the end of hospitalization (76%), and most had a good prognosis at 3 months, namely, an mRS of 0–1 (70%). After univariate analysis, seven factors were negatively associated with this good prognosis (OR < 1): age (OR = 0.90, 95% CI = 0.83–0.97, p < 0.01), confusion (OR = 0.27, 95% CI = 0.08–0.84, p = 0.02), focal deficit (OR = 0.21, 95% CI = 0.07–0.63, p = 0.01), coma (Glasgow score ≤ 8; OR = 0.05, 95% CI = 0.01–0.48, p = 0.01), deep/straight vein thrombosis (OR = 0.12, 95% CI = 0.03–0.46, p < 0.01), venous infarction (OR = 0.27, 95% CI = 0.09–0.86, p = 0.03), and cerebral hemorrhage (OR = 0.13, 95% CI = 0.04–0.46, p < 0.01; Table 2).

Comparison between male and female patients within our cohort

Male patients were significantly younger than female patients at the time of CVT occurrence (72.5 vs. 75.3, p = 0.04), and the proportion of female patients significantly increased with age (54% of patients <75 years old vs. 72% of patients aged ≥75 years old). There was no difference in the clinical pattern of symptoms, either for timing from onset to diagnosis, thrombus location, underlying pathology, or outcome between male and female patients (data not shown); we only found significantly more venous infarction among women (29% vs. 12%, p = 0.02).

Comparison between two age subgroups within our cohort

We subsequently compared patient characteristics between two age subgroups within our group: patients aged 65–74 years (n = 71) and patients aged ≥75 years (n = 43; Table 1). Symptom onset was similarly acute, subacute, and chronic in these two groups. The median time from symptom onset to diagnosis was shorter for patients of the older subgroup (2 days, IQR = 0–9 vs. 7 days, IQR = 3–21). The proportion of patients experiencing headaches as an isolated symptom was significantly lower for the older subgroup than the younger (9% vs. 24%, respectively, p = 0.04), and the proportion experiencing headaches as one of several symptoms was also significantly lower for the older subgroup (37% vs. 58%, respectively, p = 0.03); they also had more focal deficits (67% vs. 41%, p < 0.01; Table 1). Hemorrhagic complications were significantly more frequent among the older than younger subgroup, 42% versus 23% for cerebral hemorrhage (p = 0.03) and 30% versus 7% for subarachnoid hemorrhage (p < 0.01). There was no difference in associated conditions or underlying pathology. The prognosis was poorer for the older subgroup than the younger. Eight of the nine fatal outcomes in the acute phase concerned patients aged ≥75 years, making the death rate significantly higher in the older group (p < 0.01). Within our cohort, younger patients had a better outcome at 3-month and 1-year follow-up. At the end of hospitalization, the proportion of patients in the younger subgroup having a motor impairment was lower (8% vs. 30%, p = 0.01) and the proportion being able to return home was higher (86% vs. 59%, p < 0.01).

Comparison of our cohort with symptomatic CVT in adults younger than 65 years from the Lariboisière register

The proportion of female patients was lower in our cohort than adults aged <65 years from the Lariboisière register (61% vs. 79%, p < 0.01; Table 3). Symptom onset was more frequently acute for patients in our cohort (46% vs. 19%, p < 0.01). Patients aged ≥65 years were significantly less prone to experience headaches, as an isolated symptom or not, than younger adults, but the older patients presented seizures and focal deficits more frequently (p < 0.01). Coma was also a more frequent feature among younger adults (22% vs. 10%, p < 0.01). Thrombosis location involved the lateral sinus or multiple sinuses more frequently for younger adults. The only brain injury that was significantly more frequent in our cohort compared with younger adults was arteriovenous dural fistulas (5% vs. 0.6%, p < 0.01). Concerning CVT risk factors, solid cancer and hemopathy, as well as local infection (intracranial, face, or the ENT area) were significantly more frequent in patients aged ≥65 years than in younger adults (14% vs. 2% for solid cancer, 7% vs. 2% for hemopathy, and 8% vs. 3% for local infection, p < 0.01). There was no difference in the proportion of acquired and nonacquired thrombophilia between groups. The death rate in the acute phase was significantly lower for younger adults (1.7% vs. 8%, p < 0.01). We also found a significant difference when looking at the mRS at 1 year, with a poorer outcome for patients from our cohort (p < 0.01; Table 3, Figure 2).