Prediction of ShortTerm Stroke Risk After TIA

1
Prediction of Short-Term Stroke Risk After TIA

• Margaret P. Stafford, MD
• December 7, 2005

2
Introduction

• 700,000 people in the U.S. had a stroke in 2002
  (500,000 1st strokes and 200,000 recurrent
  strokes)
• 275,000 people died from stroke in 2002
• Stroke is the 3rd leading cause of death in the
  U.S. (when considered separately from other
  cardiovascular disease)
• Total cost of stroke is 57 billion per year

Statistics from American Heart Association
3
Stroke Risk After TIA

• In one study (conducted at Kaiser hospitals in
  northern CA), 10.5 of patients who presented to
  ED with a TIA had a stroke within 90 days 5 of
  patients had a stroke within 2 days (Johnston
  2000)
• In another study, stroke risk after TIA was 9.5
  at 90 days and 14.5 at one year (Hill 2004)

4
Current Practice at FHC

• How do you manage your patients who present with
  recent TIA?

5
A simple score (ABCD) to identify individuals at
high early risk of stroke after transient
ischaemic attack

• Rothwell RM, Giles MF, Flossman E, et al. Lancet
  200536629-36

6
Clinical Question

• For which patients with a recent TIA is
  emergency assessment needed, and which patients
  can be appropriately managed in a non-emergency
  outpatient setting?
• Objective derive and validate a clinical
  prediction rule to predict stroke risk within 7
  days after TIA

7
Clinical Prediction Rules

• Quantify elements of the history/physical which
  aid in diagnosis, prognosis, or treatment

From Clinical Prediction Rules,
www.usersguides.org
8
Derivation Cohort

• Population-based cohort of Oxford Community
  Stroke Project (OCSP)
• Total population 105,000 patients attending 10
  primary care practices in Oxfordshire, UK
  1981-1986
• 209 patients had TIA during study period
• 18 of these had stroke within 7 days of TIA

9
Derivation Cohort Variables Studied

• Because of small sample size, only studied
  factors already reported to be significant
  predictors of stroke after TIA
• Age
• Clinical features (motor weakness, speech
  disturbance)
• Duration of symptoms
• Diabetes
• Hypertension
• Factors with p 0.1 were included in prediction
  rule

10
Derivation Cohort ResultsPopulation
characteristics
Table 1. Characteristics of patients included in
the OCSP cohort used to derive the risk score
11
Derivation Cohort Results
Table 2. 7-day risk of stroke after presenting
TIA in relation to potential risk factors
12
Clinical Prediction Rule

• ABCD
• Age gt60 (1 point)
• Blood pressure elevation (gt140 SBP or gt90 DBP) (1
  point)
• Clinical Features unilateral weakness (2
  points), speech disturbance without unilateral
  weakness (1 point), other (0 points)
• Duration of symptoms 60 min (2 points), 10-59
  min (1 point), lt10 min (0 points)

13
Validation Cohorts

• Oxford Vascular Study dataset (OXVASC) cohort of
  90,000 patients attending 10 primary care
  practices in Oxfordshire from 2002 to 2004
• Validation Cohorts
• 1. Primary all probable or definite TIAs in
  OXVASC cohort (n190)
• 2. All suspected TIAs referred from OXVASC
  cohort (n378)
• 3. All suspected TIAs from non-OXVASC
  population of Oxfordshire referred to weekly
  hospital-based TIA clinic from 2002 to 2004
  (n206)

14
Derivation vs. Validation Cohorts Patient
Populations
Table 1. Characteristics of patients included in
the OCSP cohort used to derive the risk score and
the three other cohorts used to validate the
score

• Validation populations (in comparison with OSCP)
• More women
• More diagnosed htn and diabetes, but lower blood
  pressures

15
OXVASC Validation Cohort Results

• Of the four identified risk factors from
  derivation cohort, all except age (p0.35) were
  significant in OXVASC primary validation cohort
• Stroke risk did increase as ABCD score increased

Table 3 7-day risk of stroke stratified
according to ABCD score at first assessment in
the OXVASC validation cohort of patients with
probable or definite TIA
16
Other Validation Cohorts(all suspected TIA)
Table 5 7-day risk of stroke stratified
according to ABCD score at first assessment in
all referrals with suspected TIA to OXVASC and
risk of stroke before scheduled clinic
appointment in all referrals with suspected TIA
to the non-OXVASC hospital-referred weekly TIA
clinic

• These cohorts also showed higher stroke risk
  with higher ABCD score

17
ROC curves

• Helpful for continuous variables
• Plots sensitivity vs. 1-specificity
• Area under curve tells us how good test is (want
  at least 0.7)
• Here, area under curves for
• OXVASC suspected TIA 0.91
• OXVASC probable TIA 0.85
• Non-OXVASC suspected TIA 0.80

Figure 2. ROC curves for predictive value of ABCD
score in the three validation cohorts
18
ROC curves

• Help show tradeoff between sensitivity and
  specificity
• Choose cutoff based on whether you want greater
  sensitivity or specificity
• In this case, cutoff of gt4 as high risk, 4 as
  low risk provides a good balance of sensitivity
  and specificity

Figure 2. ROC curves for predictive value of ABCD
score in the three validation cohorts
19
Study Strengths

• Used population-based cohort more representative
  of general population
• Two of validation cohorts analyzed all suspected
  TIAs makes rule more useful to primary care
  physicians
• Predictors are easily assessed
• Rule simple, makes sense

20
Study Weaknesses

• Small sample size in all cohorts
• Could only examine previously identified risk
  factors
• Did not include time since TIA as a potential
  predictive variable
• Unclear how diagnosed stroke (apparently used WHO
  criteria)
• Does not mention whether person diagnosing stroke
  was blinded to previous TIA diagnosis or ABCD
  score
• Validation cohorts very similar to derivation
  cohort

21
Level of Evidence for CPR
Our test falls here
From Clinical Prediction Rules,
www.usersguides.org

• To improve level of evidence, we need validation
  in different population and demonstration of
  improved patient outcomes.

22
Current Management of TIA by Neurology at SFGH

• All patients with TIA within the last 1-2 days
  are admitted (high risk of stroke within first
  few days after TIA, if in-house more likely to be
  able to use TPA)
• Telemetry
• ASA, anticoagulation if indicated
• Risk factor determination and management BP,
  lipids, A1C, TTE with bubble study, CT/CTA of
  cerebral vasculature and/or carotid ultrasound
• Patients with more remote TIA usually managed as
  outpatients

23
Usefulness of ABCD Rule

• At SFGH, neurology admits all TIA patients within
  48 hours applying the ABCD rule could conflict
  with this practice
• Does this study warrant modifying the 48 hour
  rule?
• e.g., not admitting patients with ACBD score lt 4
  even if within 48 hours
• admitting patients with ABCD score of gt 4 even if
  more than 48 hours (but lt 7 days) since TIA
• In other systems the primary care or ED physician
  is in control of triage and evaluation and needs
  to decide between urgent and non-urgent follow-up
  
• Should use the 48 hour rule, the ABCD rule, or
  some combination?

24
Patient Outcomes

• Does aggressive risk factor modification after a
  TIA affect short-term stroke risk?
• Aspirin given after TIA reduces long-term stroke
  risk (Farrell 1991, Diener 1996)
• Carotid endarterectomy reduces long-term stroke
  risk in patients with TIA and high-grade stenosis
  (NASCET 1991)
• No studies look at short-term (days-weeks) stroke
  risk reduction after TIA

25
Summary

• ABCD clinical prediction rule (age, blood
  pressure, clinical features, duration of
  symptoms) for TIA is useful for predicting 7-day
  stroke risk
• In some clinical settings this rule could help
  primary care clinicians decide on immediacy of
  follow-up.

26
Citations

• American Heart Association. Heart Disease and
  Stroke Statistics 2005 Update. Dallas, Tex.
  American Heart Association 2004.
• Diener HC, Cunha L, Forbes C, Sivenius J, Smets
  P, Lowenthal A. European Stroke Prevention Study.
  2. Dipyridamole and acetylsalicylic acid in the
  secondary prevention of stroke. J Neurol Sci.
  1996 Nov143(1-2)1-13.
• Farrell B, Godwin J, Richards S, et al. The
  United Kingdom transient ischaemic attack
  (UK-TIA) aspirin trial final results. J Neurol
  Neurosurg Psychiatry. 1991 Dec54(12)1044-54.
• Hill MD, Yiannakoulias N, Jeerakathil T, et al.
  The high risk of stroke immediately after
  transient ischemic attack. A population-based
  study. Neurology 2004l 622015-20.
• Johnston SC, Gress DR, Browner WS, Sidney WS.
  Short-term prognosis after emergency department
  diagnosis of TIA. JAMA 2000 284 2901-06.
• McGinn T, Guyatt G, Wyer P, et al. Diagnosis
  Clinical Prediction Rules. InHayward, Robert,
  electronic editor. Users Guides Interactive.
  Chicago(IL) JAMA Publishing Group 2002.
  Available from http//www.usersguides.org.
• North American Symptomatic Carotid Endarterectomy
  Trial Collaborators. Beneficial effect of carotid
  endarterectomy in symptomatic patients with
  high-grade carotid stenosis. N Engl J Med. 1991
  Aug 15325(7)445-53.
• Rothwell RM, Giles MF, Flossman E, et al. A
  simple score (ABCD) to identify individuals at
  high early risk of stroke after transient
  ischaemic attack. Lancet 200536629-36.