This study has found that 74% patients initially presenting with symptoms of aphasia in the emergent setting experience full remission of their language impairments by 6 months, with partial improvement in 86% of cases. This excellent prognosis stands in stark contrast to the 40% recovery rate reported in the literature.(1
) Past studies enrolled subacute-to-chronically symptomatic subjects, usually from the population referred to speech therapy. A study of aphasia outcome by Pedersen from 1995 in which baseline assessments occurred relatively early found that half of patients with mild aphasia completely recovered within 1 week and that only 28% of patients were referred to speech therapy, whether due to medical comorbidities or resolution of symptoms.(6
) A second study on aphasia by the same group found that 19% of patients reported to have aphasia at the time of admission had experienced full resolution of symptoms by the time they could be formally evaluated by study staff at a median of 4 days after admission.(7
) These findings strongly suggest that a significant number of patients with early aphasia were not adequately captured in these early reports.
More recently, a series of studies have attempted to capture aphasia subjects within an earlier timeframe, but were limited by methodological issues which resulted in bias. One study sample was selected from patients referred for speech therapy at a mean of 11 days after stroke,(8
) another depended on speech pathology assessments at a median of 5 days from onset.(9
) A study by Lazar and colleagues was limited to first-time stroke and assessed its subjects up to 72 hours from onset.(10
) The most recent study by Inatomi, et al used NIHSS examinations to study 130 consecutive stroke patients within 48 hours of symptom onset and again at day 8-10, finding that 46% improved and 21% fully recovered.(11
The NIHSS was utilized to assess aphasia in this study as well. While the NIHSS exam is a relatively crude instrument for language assessment, it is a standardized instrument that has demonstrated reliability in the setting of emergent stroke evaluation. Furthermore, to score 0 points for language function at follow up in this study a subject would have to complete an interview with study staff in which their medical history and medications were reviewed, and a modified Rankin Scale, Barthel Index, Life Quality Scale and repeat NIHSS exam without any evidence of language impairment. This rigorous methodology provides a high level of reassurance that subjects deemed fully recovered were in fact without impairment. Use of a battery of language assessment instruments would allow for separate analysis by various aphasia features, but obtaining a baseline in the hyperacute window is not feasible. This study addresses a question pertaining to emergent management decisions in stroke where use of the standard NIHSS exam is consistent with standard practice, making our conclusions easily understood and applied to clinical practice.
It is important to note that this study continues the analytic approach established in prior studies of aphasia prognosis and excludes patients who died between interval assessments or were lost to follow up.(7
) In this study the death rate was equivalent between subjects who initially presented with aphasia and those who did not. Compared with other major studies in the literature, our data show better short term improvement rates than previously reported. Whereas the Inatomi study found that 15% of patients initially presented with aphasia, aphasia was identified in 32% of our population.(11
) This disparity may be due to the earlier assessments done in our study (median 2.3 hours and all <12 hours versus <48 hours) that would capture symptoms before resolution occurs. In Inatomi's study, only hypercholesterolemia and higher initial NIHSS score were significantly associated with early improved outcomes. In our current study, with nearly twice the number of aphasia subjects, we identified several clinical, demographic and neuroimaging characteristics predictive of improved early recovery. Regardless of those findings, the most important outcome is the long term outcome at 6 months. Studies have repeatedly shown that nearly all language recovery after stroke occurs within the first 3 months; therefore, 6 months should have been an adequate amount of time to allow for a plateau in recovery and provide a reliable measure of chronic language function.(9
) As expected, early evidence of more extensive infarction portended less early recovery. On univariate analysis this was found in a variety of markers for extensive volume of infarction: radiographically by UECT and CTA-SI, clinically by the presence of concurrent right arm weakness (a highly reliable and reproducible NIHSS exam finding) and by higher overall NIHSS score. The variables that were found to most significantly capture this phenomenon were the extent of CT hypodensity and the clinical history of prior stroke injury. For long term outcome, the presence of pre-stroke disability (by mRS) was a clear predictor of lower likelihood of improvement. Finally, a beneficial effect of smoking in terms of recovery is observed in this cohort in the univariate analysis. This apparent smoking paradox has been observed in other stroke outcome studies and is not fully understood, although the effect became insignificant when adjustment were made for other clinical variables in the logistic regression model.(13
The unique case of isolated aphasia and aphasia in patients with low total NIHSS scores is worth considering separately. These patients are often considered for thrombolysis or intra-arterial procedures on the basis that their aphasia symptoms represent a threat of significant long term disability. These data would suggest otherwise. Patients with isolated aphasia experienced complete recovery in 86% of cases (none received thrombolysis or intra-arterial treatment), and in 90% of all subjects with NIHSS<5 (p=0.02; 2 of the 41 subjects received tPA although 20 presented within 4.5 hours). Although this analysis included only cases of confirmed stroke, a recent study reviewing stroke mimics in patients who had received thrombolysis found that 3 of 11 cases (27%) of isolated aphasia were stroke mimics.(15
) Köhrmanm et al also reported on the safety and outcome after thrombolysis in stroke patients with mild symptoms (NIHSS<5) and argued that thrombolysis was appropriate in such cases because of the observed favorable outcomes in treated, versus untreated, patients with mild deficits. In their study of 32 subjects, with aphasia the most frequent symptom, 94% of subjects achieved favorable outcome with 47% fully recovering.(16
) Based on our data that included substantially more patients, only 5% treated with tPA, such outcomes would be expected by natural history alone.