Baseline Predictors of Poor Outcome in Patients Too Good to Treat With Intravenous Thrombolysis
Background and Purpose—Several studies have reported poor outcomes in patients too good to treat with intravenous thrombolysis because of mild or rapidly improving symptoms. We sought to determine baseline clinical and imaging predictors of poor outcome in these patients.
Methods—Among 3950 consecutive stroke admissions (2009–2015) in our local Get With the Guidelines–Stroke database, 632 patients presented ≤4.5 hours and did not receive tissue-type plasminogen activator, with 380 of 632 (60.1%) being too good to treat. Univariate and multivariable analyses explored the clinical and imaging features associated with poor outcome (defined as not being discharged to home) in these 380 cases.
Results—Among these 380 cases, only 68% were discharged home; the other 25% to inpatient rehabilitation, 4% to a skilled nursing facility, and 3% expired or were discharged to hospice. Patients with poor outcome were older, were more often Hispanic, had more vascular risk factors, and had higher median National Institutes of Health Stroke Scale. Imaging characteristics associated with poor outcomes included large or multifocal infarction and poor collaterals. In multivariable analysis, only age, initial National Institutes of Health Stroke Scale, and infarct location were independently associated with poor outcome.
Conclusions—Approximately one third of patients deemed too good for intravenous tissue-type plasminogen activator are unable to be discharged directly to home. Given the current safety profile of intravenous tissue-type plasminogen activator, our results suggest that the concept of being too good to treat should be re-examined with an emphasis on the features associated with poor outcome identified in our study. If replicated, these findings could be incorporated into tissue-type plasminogen activator decision-making algorithms.
Intravenous tissue-type plasminogen activator (IV tPA) remains the guideline-recommended treatment to improve outcomes after acute ischemic stroke, especially in patients without proximal arterial occlusion, and is associated with low complication rates.1,2 Unfortunately, only a small proportion of acute stroke patients actually receive tPA.3–6 Several reasons have been identified for the low rates of thrombolysis, the most common being arrival outside the appropriate time window.7,8 Up to 50% of all ischemic stroke patients initially present with mild or rapidly improving symptoms,9 and a large proportion of these patients are not treated with IV tPA, despite presenting within the 4.5-hour time window. Such patients are often considered too good to treat (TGTT) as defined by previous studies.10–12 In a previous report from our center evaluating the outcomes among TGTT patients, we found that 27% of these patients were not discharged home or died because of neurological worsening.10 Other studies have shown that up to 20% to 37% of these patients can have poor outcomes at discharge.11,13 A randomized trial of IV tPA in patients with mild stroke (PRISM [A Study of the Efficacy and Safety of Activase (Alteplase) in Patients With Mild Stroke]; unique identifier: NCT02072226) and an observational study of the nature and outcomes of mild or rapidly improving stroke (MaRISS [Mild and Rapidly Improving Stroke Study]; unique identifier: NCT02072681) are ongoing.
We sought to identify clinical and imaging factors associated with poor outcome in a modern consecutive cohort of TGTT acute ischemic stroke patients at our center.
Using our hospital’s Get With the Guidelines–Stroke (GWTG-Stroke) registry, we analyzed consecutive patients admitted with acute ischemic stroke from January 2009 to April 2015. We chose January 2009 as the start date as it reflects when we began routinely offering IV tPA in the 3- to 4.5-hour window. GWTG-Stroke is an ongoing voluntary, continuous registry and performance improvement initiative. The eligibility of each acute stroke was confirmed at chart review before abstraction. Abstracted data included patient demographics, medical history and comorbidities, in-hospital treatment and events, mortality, and discharge destination.
TGTT patients were defined as those who presented within 4.5 hours of last known well and in whom it was documented that the reason for not administering IV tPA was because stroke symptoms were too mild (NIHSS [National Institutes of Health Stroke Scale] ≤5) and rapidly improving (≥4 points improvement on NIHSS). We chose these criteria as they have been used in several previous publications to define TGTT and are currently used by the multicenter MaRISS, observational study.10–12 Imaging data collected on TGTT patients were analyzed independently by 2 trained readers who classified all images according to the following characteristics: infarct volume, infarct vessel territory (by internal carotid artery, middle cerebral artery, anterior cerebral artery, posterior cerebral artery, any cerebellar artery, other), multiple vessels in the same circulation, by circulation (anterior only, posterior only, or anterior plus posterior circulation), insular involvement, lesion laterality, adequacy of collaterals on computed tomography (CT), or magnetic resonance (MR) angiography (graded as inadequate versus adequate, with adequate collaterals being symmetrical density or signal intensity compared with the opposite side). Readers were blinded to patient characteristics and outcomes. Data were collected from any initial brain imaging, including nonenhanced CT and MR imaging, and CT or MR angiography. Not all patients underwent complete stroke imaging modality examination (CT angiography/MR imaging/MR angiography); thus, sample size of each imaging variable is reported in the Results section. Because previous literature suggests that insular infarction may influence the sympathetic or parasympathetic innervation of the heart and may predispose to cardiac arrhythmias, this area of localization was singled out for inclusion.14
Good outcome was defined as discharge to home and poor outcome as discharge to any other destination, including inpatient rehabilitation facility, skilled nursing facility, and hospice or died during the hospital stay.
TGTT patients were divided into 2 groups based on the above dichotomized outcome criteria: good outcome versus poor outcome. Student t test, χ2, and Wilcoxon rank-sum tests were used as appropriate. A 2-tailed value of P<0.05 was considered significant.
Unadjusted univariate odds ratios with 95% confidence intervals were computed for all variables, but reported for only those variables that were significant in univariate testing (P<0.1). A 2-step multivariable logistic regression was performed to identify the independent predictors of poor outcome in TGTT patients. Clinical and demographic variables were included in the first step of the regression model, and the imaging data were then added in the next step to complete the model. Given the small sample size, the arterial circulation variable was dichotomized into anterior circulation versus posterior±anterior circulation. We performed this as a 2-step analysis to distinguish between the clinical versus imaging contributions and also to allow comparability with previous studies because most have looked only at clinical characteristics and not incorporated imaging findings.
We also performed a case-matched analysis to compare the outcomes of matched patients among those who did not get tPA because of mild or rapidly improving stroke versus those who received tPA. Patients were matched based on age, then on NIHSS, 2 major factors known to be significantly associated with poor outcomes in stroke patients. Matching was carried using the SPSS FUZZY algorithm, with a tolerance of 0 for NIHSS (meaning all cases and matched control should have the exact same NIHSS) and a tolerance of 2 for age (age of the matched control should be within 2 years of the case). Statistical analyses were performed using the software SPSS, version 22.0.
A total of 3950 consecutive acute ischemic stroke were admitted from January 2009 till April 2015, with 1237 presenting within 4.5 hours of last known well. IV tPA was delivered to 605 patients (254 received IV tPA at our center and 351 received at the outside hospital and were transferred to our center for post-tPA care). Of the remaining 632 early arriving patients who did not receive IV tPA, 380 of 632 (60.1%) patients were designated TGTT by the treating physician. The majority of these patients presented with mild stroke (319/380, 84%) with median NIHSS of 2 (interquartile range, 0–4). Sixty-one patients (61/380, 16%) had rapidly improving stroke, and 55 (55/61, 90%) of these patients had NIHSS of <10 (median NIHSS of 3 [interquartile range, 1–6]). The majority (92%) of these patients presented within the 3-hour window; there were no significant differences in patient and imaging characteristics between patients presenting within 3 hours compared with those in 3-hour to 4.5-hour window.
Of the total 380 TGTT patients, only 67.8% had a good outcome (discharged home), whereas 32.2% had a poor outcome (25.1% to inpatient rehabilitation, 4.2% to a skilled nursing facility, and 2.9% expired or were discharged to hospice).
In univariate analysis of clinical factors, patients with poor outcomes were older, more often Hispanic, and had higher median NIHSS on arrival. They were less likely to be male and more likely to have a history of diabetes mellitus, coronary artery disease, and atrial fibrillation. There was no difference in the rates of hyperlipidemia, previous stroke, previous transient ischemic attack, and smoking between patients with good versus poor outcome (Table 1).
In univariate analysis of imaging variables, patients with poor outcome were more likely to have infarcts that involved both hemispheres and the posterior (with or without anterior) circulation. Not unexpectedly, patients with poor outcomes had larger diffusion-weighted imaging lesion volumes on admission. There were no differences in terms of the specific single-vessel territory involved or the proportion of patients with insular infarction between patients with good versus poor outcome (Table 2).
Patients with poor outcomes were more often discharged to inpatient rehabilitation facility (78.7%), then to skilled nursing facility (12.3%), and 9.0% expired or were discharge to hospice care. There was no significant difference in the modified Rankin Scale before admission between patients with good versus poor outcome, and all but 2 patients were independent in ambulation before the presenting stroke. At the time of discharge, all patients with good outcome were independent in ambulation. Length of stay was longer in patients with poor outcome (Table 3).
In the first step of this 2-step multivariable analysis where we adjusted only for clinical factors, age, Hispanic ethnicity, and initial NIHSS remained significantly associated with poor outcomes. In the second step of the model when adjusting for both clinical and radiological factors, age and Hispanic ethnicity remained significant, and diabetes mellitus and involvement of the posterior (with or without anterior) circulation were also independently associated with poor outcome (Table 4).
On case-match analysis, the matching algorithm returned 222 cases in each group (TGTT versus tPA treated), with a median NIHSS of 2 (1–4) in both groups and mean age of 68.3±13.1 in TGTT versus 68.2±13.1 in tPA-treated patients. When comparing outcomes, 37.4% (83/222) of patients who were TGTT had poor outcomes compared with 29.7% (66/222) of patients who received tPA (P=0.05).
Even at an academic medical center with a predisposition to provide IV thrombolysis, mild or rapidly improving stroke is an exceptionally common reason why patients eligible by time and other medical criteria are not treated with IV tPA. The exclusion of these patients from thrombolytic therapy is based on the assumption that if left untreated they will have an excellent functional outcome, thus not justifying the potential risks of reperfusion therapy. The inclusion criteria in the NINDS (National Institute of Neurological Disorders and Stroke) tPA trials required a neurological deficit measurable on the NIHSS without a prespecified minimum value, and the range of NIHSS values was 1 to 37.4 If one assumes that IV tPA use would only be withheld in the case of a nondisabling minor deficit, then one would expect that the vast majority of these TGTT patients would be discharged directly back to home. By contrast, we found that a substantial number of TGTT patients (32%) were unable to be discharged home based on the degree of their neurological disability and required further institutional care or died. We found several clinical and imaging stroke characteristics, which were independently associated with poor outcomes in these patients. These findings support previous work, suggesting less than ideal short-term outcomes in these patients.
Being considered TGTT is a common reason for not administering IV tPA in eligible patients. In the NINDS study, ≈13% of screened but not enrolled patients had experienced rapid improvement of symptoms.4 In a series of acute stroke patients from Calgary, 13% had mild symptoms and 18% had rapidly improving symptoms, whereas in a 9-hospital system study from Cleveland, mild symptoms were listed as the reason for no tPA in 77% and rapidly improving symptoms in 44% of those arriving <3 hours, but the same patient could have been excluded for both reasons.15 Similarly, a population-based study from Ohio and Kentucky found that mild symptoms accounted for 43% of exclusions of <3-hour patients from reperfusion therapy.16,17 Our previously published report showed that being considered TGTT was the most common reason for exclusion from IV tPA treatment at our center (58%).10 Although these previous studies included only patients considered in the 3-hour window of eligibility for IV tPA, our current study extends the analysis to the group of patient presenting up to 4.5 hours. Most importantly, our current study is the first of its kind evaluating outcomes among TGTT patients to include detailed imaging information, which significantly modifies the associations of previously reported variables, such as initial stroke severity.
Our study confirms reports of earlier studies that the natural history of patients with mild or rapidly improving symptoms may not follow its expected benign course. A substantial number of patients were not discharged home, and in ≈10% of patients the disability during the course of hospital stay was significant enough that they warranted a discharge to skilled nursing facility or were terminally ill (death or hospice care). These results are supported by other similar studies available in literature.18 In our previous study, patients with rapidly improving symptoms were 4 times more likely to experience subsequent neurological deterioration. This is consistent with a previous report by Rajajee et al19 who observed that the clinical course of stroke fluctuates more in the first 3 hours than in hours 3 to 6, so that improved or mild deficits are more reliably predictive of good final outcome when they are observed in the 3- to 6-hour than in the 0- to 3-hour window.
By using clinical, demographic, and initial MR parenchymal imaging and CT or MR vascular imaging information, we can potentially identify patients who are initially considered TGTT but who are more likely to have a poor outcome without IV tPA. Although we do not know whether these TGTT patients would have benefited from IV tPA, our data support the rationale for the current randomized trials of these TGTT stroke patients. MaRISS is an observational clinical trial nested inside GWTG-Stroke and will potentially provide additional information about the natural history of a multicenter prospectively acquired cohort of these patients.
Our study identified important variables that were associated with poor outcome in univariate testing, and interestingly, several of these factors did not remain significant in multivariable testing. Coronary artery disease and atrial fibrillation, which are often markers of larger more disabling strokes, are often predictors of poor outcome after stroke. However, after adjustment for other clinical variables, these 2 were no longer significantly associated with outcome, likely because of the much stronger predictive variable of NIHSS. The NIHSS has been previously shown to be the most potent predictor of inpatient mortality in GWTG-Stroke, and it is likely that in this mildly disabled cohort the impact of atrial fibrillation and coronary disease is muted.20,21 Advanced age and diabetes mellitus both increased the likelihood of a delayed discharge to home, and this effect was largely unchanged by multivariable regression. Both have been previously shown to negatively impact outcomes. Interestingly, the association of Hispanic ethnicity and poor outcome increased with adjustment. Because there is no obvious biological explanation for this effect, it suggests that this and other unmeasured social determinants of health may be influencing the discharge destination.
Among the imaging variables, many were significant in univariate testing, but none remained significant except for posterior circulation involvement. This was despite large odds ratios and highly significant P values. In considering this finding, it is important to remember that this cohort is restricted to mild strokes and that the other candidate imaging variables are all factors, which have been shown to influence outcomes in strokes in cohorts with a much broader range of initial severity. Although there was a small percentage of strokes with NIHSS >8 suggestive of a proximal occlusion where collateral supply will be important, or of cardiac or proximal source embolism affecting multiple vascular territories or circulations, more than half of the cases with poor outcome had an initial NIHSS of 0 to 2, and 75% had an initial NIHSS of ≤4. Because the NIHSS does not measure swallowing function, urinary retention, or gait instability, it is possible that these clinical phenotypes influenced outcome and are unmeasured intermediate outcomes that are strongly associated with being unable to be discharged to home. With a median admission diffusion-weighted imaging lesion volume of 2.1 mL, it is also unlikely that the volume itself over and above location and NIHSS has a chance to exert its usual influence in this mild cohort. It is also possible that other measured or unmeasured clinical variables are exerting an influence over the outcome that is independent of the imaging findings.
Last, it is important to note that we do not have direct evidence that treating the TGTT patients with tPA would have improved their outcomes. This can only be answered with a randomized controlled trial, such as PRISMS. However, our study confirms the unexpectedly high rate of poor outcomes in the TGTT population and provides supporting evidence in our case-matched analysis that tPA may be of benefit in these patients.
Our study has several important limitations worth noting. We included all consecutive ischemic stroke admissions to our hospital; however, there is a strong potential for selection bias at the level of the treating team and attending stroke neurologist in deciding which mild or rapidly improving patients received IV tPA and which did not. Because our data are limited to one center, larger multicenter studies should be performed to verify our results and make them more generalizable to nationwide hospitals. In addition, the majority of our TGTT patients arrived within the 3-hour time window; thus, further studies are needed to explore whether the factors associated with poor outcomes are different among those presenting in the extended window of 3 to 4.5 hours. We do not have longer-term outcomes for our cohort, and it is likely that many patients who were dependent at hospital discharge recovered independence by 90 days; however, being able to be discharged directly to home and time spent at home during the first 3 months after stroke is an important outcome and one that has been identified as a patient-reported outcome measure highly prioritized by patients and their families.22 We do not have a clear explanation for why patients of Hispanic ethnicity were less likely to be discharged home. It is possible that social determinants of health, such as health literacy, family income, cultural beliefs, and the availability of a caregiver at home, may be playing a role in determining discharge destination. Finally, in-hospital complications can also influence the discharge destination and prevent home discharge. Although the 2 cohorts had similar performance on the GWTG-Stroke achievement measures (data not shown), it is possible that a reduction in disability because of tPA treatment could decrease the risk of subsequent complications and thereby improve discharge outcome.
A substantial percentage of patients deemed TGTT with IV tPA has residual deficits that prevent discharge directly to home. The practice of administering IV tPA to patients with mild or rapidly improving deficits has been demonstrated to be quite safe, with low symptomatic intracranial hemorrhage rates and good outcomes. Pending the results of the PRISMS trial, when considering withholding treatment for patients with mild or rapidly improving symptoms, it may be prudent to urgently evaluate other functional domains that could prevent home discharge, such as gait and swallowing, and to obtain urgent MR imaging to determine whether tPA treatment is warranted. Given how common TGTT is as a reason for not administering IV tPA and that outcomes among these patients are still in question, prospective multicenter trials are needed to ascertain whether the findings in our current study are generalizable; that is, to determine whether advanced age, Hispanic ethnicity, higher initial NIHSS, and infarction involving the posterior circulation are associated with worse outcomes in untreated patients. In addition, further research is needed to understand the frequently unmeasured social determinants of health that prevent home discharge in some patient populations and to eradicate these differences when present.
Dr Schwamm reports being the principal investigator of an investigator-initiated study of extended-window intravenous thrombolysis funded by the NINDS (clinicaltrials.gov/show/NCT01282242) for which Genentech provides alteplase free of charge to the Massachusetts General Hospital, as well as supplemental per-patient payments to participating sites; serving as chair of the American Heart Association/American Stroke Association Get With the Guidelines-stroke clinical work group; serving as a stroke systems consultant to the Massachusetts Department of Public Health; and serving as a scientific consultant regarding trial design and conduct to Lundbeck (International Steering Committee; DIAS-3, DIAS-4 trials [Safety and Efficacy of Desmoteplase Given 3–9 h After Ischaemic Stroke in Patients With Occlusion or High-Grade Stenosis in Major Cerebral Arteries]), Penumbra (Data and Safety Monitoring Committee, Separator 3D trial), and Medtronic (Victory AF [Atrial Fibrillation], REACT AF, and Stroke AF trials). The other authors report no conflicts.
Continuing medical education (CME) credit is available for this article. Go to http://cme.ahajournals.org to take the quiz.
Guest Editor for this article was Giuseppe Lanzino, MD.
- Received July 22, 2016.
- Revision received September 13, 2016.
- Accepted October 5, 2016.
- © 2016 American Heart Association, Inc.
- Hacke W,
- Donnan G,
- Fieschi C,
- Kaste M,
- von Kummer R,
- Broderick JP,
- et al
- Lansberg MG,
- Bluhmki E,
- Thijs VN
- Chapman SN,
- Mehndiratta P,
- Johansen MC,
- McMurry TL,
- Johnston KC,
- Southerland AM
- Yu AY,
- Hill MD,
- Coutts SB
- Smith EE,
- Abdullah AR,
- Petkovska I,
- Rosenthal E,
- Koroshetz WJ,
- Schwamm LH
- Willey JZ,
- Stillman J,
- Rivolta JA,
- Vieira J,
- Doyle MM,
- Linares G,
- et al
- Yu AY,
- Hill MD,
- Coutts SB
- Colivicchi F,
- Bassi A,
- Santini M,
- Caltagirone C
- Katzan IL,
- Sila CA,
- Furlan AJ
- Khatri P,
- Kleindorfer DO,
- Yeatts SD,
- Saver JL,
- Levine SR,
- Lyden PD,
- et al
- Balucani C,
- Levine SR
- Smith EE,
- Shobha N,
- Dai D,
- Olson DM,
- Reeves MJ,
- Saver JL,
- et al
- Xian Y,
- O’Brien EC,
- Fonarow GC,
- Olson DM,
- Schwamm LH,
- Hannah D,
- et al