Factors Associated With the Decision to Hospitalize Patients After Transient Ischemic Attack Before Publication of Prediction Rules
Background and Purpose— One important criterion for hospitalizing patients after transient ischemic attack (TIA) is the short-term risk of stroke. Before publication of prediction rules for stroke after TIA, physician judgment was required to make a decision about hospitalization. We sought to identify factors associated with the decision to admit patients with TIA from the emergency department (ED) and to see whether those at highest risk of stroke were selected for admission.
Methods— All patients diagnosed with TIA in the ED of 16 hospitals in the Kaiser-Permanente Medical Care Plan over a 1-year period before publication of prediction rules were included (n=1707). Risk of subsequent stroke was stratified according to a validated prediction rule (ABCD2 score), and the decision to admit was correlated with these risk scores. Factors associated with admission in univariate analysis were included in a logistic regression model.
Results— Overall, 243 patients with TIA (14%) were admitted. Admission weakly correlated with the ABCD2 score (rank biserial R2=0.036; 10.0% at low 2-day risk of stroke admitted versus 20.3% at high risk). Seven variables were independently associated with a decision to admit after TIA: prior TIA, speech impairment, weakness, gait disturbance, history of atrial fibrillation, symptoms on arrival to ED, and use of ticlopidine.
Conclusions— In this cohort of patients with TIA, the decision to admit was weakly correlated with risk of subsequent stroke as measured by the ABCD2 score, and several risk factors for stroke were not important for the decision to admit. Before publication of prediction rules for stroke after TIA, physicians were not identifying the majority of patients at highest risk of stroke for admission.
Consensus guidelines suggest that clinicians should consider hospitalizing patients after transient ischemic attack (TIA), especially when it has occurred in the prior 24 to 48 hours.1 However, practices regarding admission after TIA vary widely, with a national admission rate of just over 50% that has not increased in the past decade and varies substantially by geographic region.2
Numerous criteria have been proposed to determine whether TIA patients should be hospitalized, including the availability of rapid outpatient workup or neurologic consultation, the timing of the spell in relation to presentation, the presence of critical carotid stenosis requiring urgent surgery, and the presence of other unstable medical conditions.1,3–5 One other potentially important criterion for admission is the short-term risk of subsequent stroke after TIA; patients at low risk for subsequent stroke likely do not need inpatient observation, whereas those at high risk may benefit from expedited workup, prompt initiation of secondary prevention, and rapid access to acute treatments such as thrombolysis should subsequent events occur.6,7
In the last decade, multiple population-based studies have demonstrated that the short-term risk of subsequent stroke after TIA is very high, particularly in the first few days after the event.8–10 The California score was proposed as a simple tool for stratifying stroke risk after TIA, and this was subsequently modified and extended to create the ABCD score.8,9 A recently published unified score allows for estimation of 2-day risk of stroke after TIA using 5 variables including age, blood pressure, clinical TIA symptoms, duration of spell, and presence of diabetes (“ABCD2”).10 Before the publication of risk stratification models, physician judgment alone was used to determine whether admission was indicated after TIA.
We aimed in this study to identify factors associated with the decision to admit patients presenting to the emergency department (ED) with TIA before the publication of these risk models. We compared factors associated with admission to the validated risk score because using actual rates of stroke in those admitted would likely be confounded by potential benefits of admission. Investigating practices before the publication of these risk scores is key in determining the importance of these models: if physician judgment regarding stroke risk was already accurate and impacting the decision to admit, then wide acceptance of these validated risk scores will have little impact on TIA care.
All patients diagnosed with TIA in the ED of 16 hospitals in the Kaiser-Permanente Medical Care Plan over a 1-year period ending February 1998 (before publication of prediction rules) were included (n=1707), as previously detailed.8 The diagnosis of TIA was assigned directly by the ED physician, not by the treating physician during the hospitalization. Characteristics of the patient and TIA were abstracted, including demographics, past medical history, medications taken at the time of TIA, TIA symptoms, and the results of neurologic and general physical examinations. Patients were followed for subsequent stroke within the 90 days after presentation. Strokes were confirmed by independent adjudication of events by 2 neurologists. A small number of patients were excluded (n=20) who, during the same ED visit for TIA, had a subsequent stroke prompting admission to the hospital.
ABCD2 score was calculated for each patient. The decision to admit patients was correlated with ABCD2 score using rank biserial correlation (R2) and the Cochrane Armitage test. In addition, factors associated with admission in univariate analysis (P<0.20) by Fisher exact test were included in a logistic regression model, with subsequent backward stepwise elimination of variables no longer significant (P>0.10). A Cochran-Armitage test for significance was used to compare admission rates between low-, medium-, and high-risk groups by ABCD2 score. All statistical calculations were made using SAS (Version 9.1).
Overall, 243 of 1687 TIA patients were admitted to the hospital (14%). The decision to admit only weakly correlated with the ABCD2 score, with less than 4% of variance in the decision to admit explained by the risk score (rank biserial R2=0.036, P=0.0001; Cochrane Armatage, P=0.0001). Using ABCD2 risk stratification, 10.0% of patients with low risk (score 0 to 3) of subsequent stroke in 48 hours were admitted compared with 20.3% of patients at high risk (score 6 to 7) for subsequent stroke (Figure). These differences in the rates of admission between risk stratification groups were significant (P<0.0001).
In univariate analysis, several factors were associated with the decision to admit (Table 1). In multivariable models, 7 variables were found to be independently associated with a decision to admit after TIA: atrial fibrillation; prior TIA; TIA symptoms persisting on arrival to the ED; use of ticlopidine at the time of the spell; and speech impairment, gait disturbance, or weakness as a clinical feature of the TIA (Table 2).
In this cohort of patients with TIA, the decision to admit to the hospital only weakly correlated with the short-term risk of subsequent stroke before the publication of validated risk stratification models. Several important identified risk factors for subsequent stroke, including age, blood pressure, and a history of diabetes, did not impact the decision to admit. Many TIA patients were not admitted to the hospital before publication of these risk models as the short-term risk of subsequent stroke was unknown.
Some factors associated with admission may have been appropriate for reasons other than short-term stroke risk. For example, atrial fibrillation was found to be independently associated with a decision to admit patients with TIA; although atrial fibrillation has not been associated with an increased risk of short-term stroke,8 some of these patients may have been admitted for rapid cardiology consultation, rate control, or initiation of anticoagulation.11 Concern about persistence of symptoms may also have influenced decision making as well as a possible need for physical or speech therapy. Prior TIA on presentation may have been interpreted as failure of prior therapy, prompting additional concern. It should be noted, however, that none of these factors have been associated with increased risk of subsequent stroke, though prior history of TIA is associated with a greater risk of recurrent TIA.12
Specific selection criteria for short-term inpatient workup and monitoring in patients with TIA remain controversial. However, the decision likely should be based in part on the individual patient’s short-term risk of subsequent stroke. The ability to initiate thrombolysis more frequently and rapidly alone may justify hospitalization, but only for patients with higher short-term risk of stroke.7 Before the publication of validated risk stratification models, ED physicians were not identifying those TIA patients at highest risk for stroke. This finding emphasizes the clinical utility of these prediction models and argues for their widespread dissemination among neurologists, ED physicians, and other health care workers evaluating patients after TIA. Planned future studies examining admission practices after the publication of these prediction rules will be important to gauge their impact on clinical practice, including admission rate and outcomes.
- Received April 16, 2007.
- Revision received June 27, 2007.
- Accepted July 16, 2007.
Johnston SC, Nguyen-Huynh MN, Schwarz ME, Fuller K, Williams CE, Josephson SA, Hankey GJ, Hart RG, Levine SR, Biller J, Brown RD Jr, Sacco RL, Kappelle LJ, Koudstaal PJ, Bogousslavsky J, Caplan LR, van Gijn J, Algra A, Rothwell PM, Adams HP, Albers GW. National Stroke Association guidelines for the management of transient ischemic attacks. Ann Neurol. 2006; 60: 301–313.
Donnan GA, Davis SM, Hill MD, Gladstone DJ. Patients with transient ischemic attack or minor stroke should be admitted to hospital: for. Stroke. 2006; 37: 1137–1138.
Lindley RI. Patients with transient ischemic attack do not need to be admitted to hospital for urgent evaluation and treatment: against. Stroke. 2006; 37: 1139–1140.
Ovbiagele B, Saver JL, Fredieu A, Suzuki S, Selco S, Rajajee V, McNair N, Razinia T, Kidwell CS. In-hospital initiation of secondary stroke prevention therapies yields high rates of adherence at follow-up. Stroke. 2004; 35: 2879–2883.
Nguyen-Huynh MN, Johnston SC. Is hospitalization after TIA cost-effective on the basis of treatment with tPA? Neurology. 2005; 65: 1799–1801.
Santini M, De Ferrari GM, Pandozi C, Alboni P, Capucci A, Disertori M, Gaita F, Lombardi F, Maggioni AP, Mugelli A, Salerno-Uriarte JA, Sermasi S, Schwartz PJ, FIRE Investigators. Atrial fibrillation requiring urgent medical care. Approach and outcome in the various departments of admission. Data from the atrial Fibrillation/flutter Italian REgistry (FIRE). Ital Heart J. 2004; 5: 205–213.
Johnston SC, Sidney S, Bernstein AL, Gress DR. A comparison of risk factors for recurrent TIA and stroke in patients diagnosed with TIA. Neurology. 2003; 60: 280–285.