Mortality in a Multiethnic Population Attending a One-Stop TIA Clinic

Introduction: Studies indicate a 13–27% mortality rate following a transient ischaemic attack (TIA). However, outcomes following TIA/minor stroke since the introduction of rapid-access TIA clinics and prompt vascular risk factor intervention are not known. Specifically, there is paucity of data comparing outcomes between people who are diagnosed with an “acute cerebrovascular” (CV) event or an alternative non-cardiovascular diagnosis (non-CV) in a rapid-access TIA clinic. We aimed to assess the mortality in such a setting. Methods: A retrospective observational study was undertaken at the Leicester rapid-access secondary care TIA clinic. Data included information collected at the first clinic visit (including comorbidities, and primary diagnosis, categorized as CV and non-CV) and the date of death for people dying during follow-up. Results: 11,524 subjects were included with 33,164 years of follow-up data; 4,746 (41.2%) received a CV diagnosis. The median follow-up time was 2.75 years (interquartile range 1.36–4.32). The crude mortality rate was 37.3 (95% CI: 35.3–39.5) per 1,000 person-years (PTPY). The mortality rate was higher following a CV diagnosis (50.8 [47.2–54.7] PTPY) compared to a non-CV diagnosis (27.9 [25.7–30.4] PTPY), and for males, older people, those of white ethnicity, and people with orthostatic hypotension (OH). Discussion: This study identified possible risk factors associated with a higher mortality in TIA clinic attendees, who may benefit from specific intervention. Future research should explore the underlying causes and the effect of specific targeted management strategies.


Introduction
Transient ischaemic attack (TIA) is defined as a transient episode of neurological dysfunction caused by focal brain, spinal cord, or retinal ischaemia, without acute infarction [1]. TIA, reversible ischaemic neurological deficit, and minor stroke represent a continuum of cerebrovascular disease rather than distinct subtypes [2], and people who have had a TIA have a high early stroke risk [3]. Follow-up of the EXPRESS study found that early initiation of treatment following TIA or minor stroke was associated with a substantial long-term reduction in recurrent stroke [4]. This led to the establishment of rapidaccess TIA clinics across the UK, providing urgent assessment and treatment of people with suspected TIA.
The OXVASC study reported 5-year death or disability rates of 70% after a stroke and 48% following a TIA [5]. Several studies have sought to ascertain risk factors for mortality and morbidity following stroke and TIA. A variety of factors have been previously associated such as age, sex, previous TIA or stroke, atrial fibrillation (AF), ethnicity, systolic blood pressure (BP) [6][7][8]. Given the significant mortality and morbidity of TIA and stroke, understanding the characteristics of those at the highest risk of poor long-term outcome is imperative. Furthermore, a large proportion of people turn out to have TIA mimics [9], with limited data on their long-term outcomes. However, even non-focal transient neurological symptoms are associated with a higher risk of stroke and dementia [10]. Therefore, we aimed to assess risk of mortality of people seen in a contemporary single centre rapid-access, one-stop TIA clinic.

Study Design and Data Source
This retrospective observational study for quality improvement evaluation of the Leicester rapid-access TIA clinic was part of the Leicester Stroke and TIA Research (LeiSTAR) study, including people attending the Leicester one-stop TIA clinic. Ethical approval was provided by the Cambridge Central REC (April 19, 2018; REC: 17/ EE/0412) and regulatory approval by the University Hospitals of Leicester NHS Trust. The data included information collected at the first clinic visit on the subject's age and date of death, for those dying during the follow-up period. Mortality status was acquired from the hospital mortality status. This captures information about death status from hospital records and the NHS Spine, which covers everyone registered with a general practitioner in England and Wales. Subjects without BP information or with one or more BP measures of zero were excluded from the analysis. The follow-up period began on the date of the first clinic visit (earliest September 28, 2012). The end of the study period was March 16, 2019.
BP measures (seated and standing) were obtained by a trained healthcare professional at a clinic review. These measures were examined continuously and as 10 mm Hg units for hazard ratios. Orthostatic hypotension (OH) was defined as a reduction in systolic BP ≥20 mm Hg or a reduction in diastolic BP ≥10 mm Hg, within 3 min of standing [11].

Outcome
The outcome of interest was time until death.

Covariates
Sex and age were obtained from clinic records. Ethnicity was self-reported and categorized as white, black, South Asian, other, or unknown. Pre-existing AF was determined from either "atrial fibrillation" or "AF" listed in the free text for relevant past medical history or listing of apixaban, dabigatran, rivaroxaban, or warfarin as current medication. Pre-existing hypertension was determined from either "high blood pressure," "HTN," or "hypertension" listed in the free text for relevant past medical history. Pre-existing TIA and/or stroke was determined from either "TIA," "Transient ischaemic attack," or "stroke" listed in the free text for relevant past medical history. Pre-existing diabetes was determined from either "T2DM," "diabetes," "DM," "type 2," "insulin," "diet controlled," or "hyperglyc*" listed in the free text for relevant past medical history. Pre-existing hypercholesterolaemia was determined from either "high cholester," "hypercholester," "hyperlipid," "dyslipid" or "high lipid" listed in the free text for relevant past medical history, and reviewed manually. Smoking was categorized from patient self-report of status at event as current, ex, and never. Excessive alcohol consumption was determined by subjects self-reporting more than 14 units of alcohol consumed per week. CV diagnoses included the following principal diagnoses made at the first rapidaccess clinic appointment: transient ischaemic attack, ischaemic stroke, transient monocular blindness, retinal artery occlusion, intracerebral haemorrhage, and subarachnoid haemorrhage.

Statistical Analyses
The mortality rate and 95% confidence interval (CI) were calculated in total and by primary diagnostic category (CV/non-CV) at TIA clinic visit, stratified by age group, sex, ethnicity, and presence of OH. Kaplan-Meier estimates were used to measure the pro-portion of subjects that survived for up to 6.5 years following their first rapid-access clinic appointment, depending on the primary diagnostic category, stratified by sex and presence of OH. Cox regression modelling was used to investigate the association between seated and standing BP measures, as well as the presence of systolic and diastolic OH and time until death. Crude and adjusted hazard ratios and 95% CIs were calculated for seated and standing BPs and systolic and diastolic OH, and the adjusted hazard ratios were plotted stratified by primary diagnostic category. The model was adjusted for sex, age, ethnicity, previous AF diagnosis, previous hypertension diagnosis, previous TIA/stroke diagnosis, previous diabetes diagnosis, and primary diagnostic category. Initial analyses were performed in STATA 15 and SAS v9.4. Further analysis undertaken was performed in SPSS v28. Statistical comparisons of cohort characteristics were carried out using χ 2 , independent t test and Mann Whitney U tests as appropriate.  diagnosis; TIAs were diagnosed in 2,323 (20.2%), and ischaemic stroke in 1,855 subjects (16.1%). 6,778 subjects received a non-CV diagnosis. The most common non-CV diagnoses were no formal diagnosis (21.8%), migraine (13.4%), and other (9.7%). The prevalence of common vascular risk factors by primary diagnostic category is shown in Table 1.

Mortality and Diagnosis
Crude mortality in the CV group was 14.6%, compared to 8.0% in the non-CV group, with most deaths occurring in the first 3 years (  Table 3).

Mortality and Sex
Crude survival by diagnosis stratified by sex is shown in Figure 1. Females were more likely to survive compared to males, regardless of diagnosis.
A CV diagnosis (compared to non-CV) was associated with higher mortality, and a more pronounced effect was noted in females (rate ratio 2.18, p < 0.001), than males (rate ratio 1.52, p < 0.001). Mortality amongst males was significantly higher (33.9 PTPY, vs. females 23.0 PTPY; rate ratio 1.48, p < 0.001) in the non-CV cohort but not the CV cohort.
Kaplan-Meier estimates of survival by primary diagnostic category in the oldest age group (85 years and older), stratified by sex, are shown in Figure 2. In males aged 85 years and older, those with a non-CV diagnosis were less likely to survive.

Mortality and BP
The mortality rate was higher in people with either systolic or diastolic OH regardless of the primary diagnostic category (Table 3). Online supplementary Figure II shows adjusted Cox proportional hazard ratios for all-cause mortality by seated BP, stratified by diagnosis. In subjects with a CV diagnosis, lower systolic and diastolic seated pressures were associated with a higher hazard of mortality. A similar result was seen with standing BP (online suppl. Fig. III) and was particularly notable with diastolic pressure.

Discussion
This observational study from a large secondary care centre highlights the profile of those at high-risk of mortality in a TIA clinic. Vascular diagnosis, older age, male sex, WE ethnicity, lower seated SBP and DBP, and OH were associated with increased mortality. However, in males over 85 years of age, non-CV diagnosis (vs. CV diagnosis) was associated with worse outcomes.

TIA Clinic Non-CV Diagnoses
Overall, nearly 60% of people received a non-CV diagnosis, in keeping with other studies where 50-60% of suspected TIA patients were found to be mimics (non-vascular diagnoses) on further evaluation [9,12]. This may relate to differences in knowledge and skills of those referring to the clinic [13], difficulty in diagnostic validity of focal neurological symptoms, and a tendency to cautious over-referral.

Mortality
The overall mortality rate for all people seen in the TIA clinic was 37.3 PTPY. During the entire follow-up period of this study, 14.6% of people with an acute CV diagnosis died, which was in keeping with the mortality of 12.8% for those with a TIA reported by Dutta et al. [9] at 4-year follow-up. The mortality amongst people with a CV diag- nosis seen in our study was considerably lower than the OXVASC study, where 27% of those with a TIA had died by end of the 5-year follow-up period [5]. This may reflect recent changes in clinical practice with the introduction of rapid-access TIA clinics, and earlier introduction of aggressive secondary prevention therapies, including dual antithrombotic therapy, statins, BP-lowering, and secondary prevention of AF. Similar to Dutta et al. [9], we believe that limitations with our methodology are unlikely to explain the difference in mortality. It seems unlikely that non-vascular diagnoses are being misdiagnosed as CV disease considering that nearly 60% of people who attended clinic did not receive a CV diagnosis. The higher mortality rate amongst people with a CV diagnosis is in keeping with the established association between vascular risk factors and increased mortality. However, a significant mortality of 8.0% was noted in the non-CV group. These findings are comparable to the 8.0% mortality amongst those receiving non-vascular diagnoses in the Dutta et al. [9] study. Whilst overall mor-tality following non-CV diagnoses is lower compared to CV diagnoses, further study is warranted to explore associated factors.

Risk Factors
Mortality rate was higher in males compared to females, in both the CV and non-CV groups. This could be due to several reasons, for example age of presentation to the clinic. The sex disparity needs confirmation in other studies, and further research is warranted to identify the associated factors.
A higher mortality rate was seen in the CV group, compared to the non-CV group, in both men and women. In women, the difference in mortality rate was particularly pronounced. Recognized disparities in management of women with ischaemic stroke provide a potential explanation, for instance Carcel et al. [14] found that women with ischaemic stroke had lower odds of having an antiplatelet, glucose lowering therapy, and lipid lowering therapy on admission to hospital. In males over 85, those with a non-CV diagnosis had higher mortality. However, this was only seen towards the end of followup, and statistical interpretation was limited by small numbers. Mortality rate increased in both groups with increasing age, particularly when looking at those 85 years and older. Studies have noted that older people with stroke have a worse prognosis [15], particularly in those aged 85 years and older [16]. It is important to note that Kammersgaard et al. [17] found that other factors were also independent risk factors for long-term survival such as initial stroke severity and AF, and some very old people are likely to benefit from treatment. Considering other risk factors is particularly pressing in this group, as well as taking a holistic point of view of further management.
The presence of systolic or diastolic OH increased mortality rates in both people with a CV diagnosis and non-CV diagnosis [18]. Meta-analysis data have shown that OH is associated with 40% higher risk of future allcause mortality events [19]. In subjects with a CV diagnosis, lower seated and standing systolic and diastolic BPs were associated with higher hazard ratio for mortality. Lin et al. [7] demonstrated that low to normal systolic BP was associated with poorer outcomes in patients with a history of stroke. The CV diagnosis group included older people with history of AF and hypertension, who may have some degree of heart failure, which may contribute to poorer outcome with lower BP. In people over 85 years of age on antihypertensive therapy, it has been shown that lower systolic BP is associated with higher mortality and faster decline in cognitive function [20]. Particularly in older adults, it may be prudent to weigh up the benefits and risks of intensive BP lowering.

Strengths and Limitations
This study included 11,524 people who presented to a rapid-access TIA clinic. The large cohort is diverse in terms of age and ethnicity and the clinic represents a typical rapid-access TIA clinic, with similar practices used across the UK. Diagnosis was made by a variety of clinicians, as part of routine clinical practice as opposed to a research setting, thus making results representative of current clinical practice, at the cost of some lack of standardization.
It is well recognized that acute cerebrovascular disease, including both TIA and minor stroke, is associated with disability [21]. We were unable to include disability as an outcome, as these data were unavailable. We used a retrospective, observational study design. Therefore, the findings are subject to the inherent limitations of all observational studies, primarily that causality cannot be inferred, and that our analysis would not have included all relevant prognostic factors. It was not possible to blind data collectors to the diagnosis made in TIA clinic, but both the exposures of interest and outcomes were well defined to minimize the risk of bias. Unfortunately, we did not have robust, complete data on risk factors such as BMI and other lifestyle factors to incorporate these into the analysis. Lack of data on prior cardiac conditions may have affected the blood pressure and OH analyses.
A comparison of mortality before and after introduction of rapid-access TIA clinics would have provided important external validation of the OXVASC data with regards to the impact of such clinics, however we do not have historical data to undertake such a comparison. This work has been undertaken before [22].

Future Direction
Understanding why certain population groups are at higher risk of mortality may help to develop novel management strategies. As such, further research into the sex and ethnic differences, and causes of death in this population is required. Furthermore, there should be a focus on understanding the mechanism by which OH may lead to increased mortality [18].

Conclusion
This study identifies risk factors for increased mortality in the TIA clinic, including: CV diagnosis, older age, WE ethnicity, and OH. Future research should look to understand the reasons behind this and develop effective management strategies to lower risk in these groups.

Data Availability Statement
Data generated or analysed during this study are included in this article and/or its online supplementary material files. Further enquiries can be directed to the corresponding author.