|dc.description.abstract||In 2015, 11000 patients were treated in Norwegian hospitals with stroke as their primary diagnosis. This number may increase substantially over the next decades due to more elderly and very old people in the population. In most stroke patients, the symptoms are more severe on onset. However, some patients have a clinical course where the symptoms increase after onset, which is now most commonly termed early neurological deterioration (END). Plausible mechanisms for END after ischemic stroke have been presented. However, the predictors identified vary between studies, and are partly contradictory. In studies, the methods to detect END differ both in regard to when and how often to observe patients to identify END, and which stroke scales are used. Even when using the same stroke scale, the thresholds used differ between studies. Recently, NIHSS has been the dominating stroke scale for END detection.
This thesis had two important aims. The first main aim was to improve the methods for detection of END. Therefore, in order to evaluate whether frequently repeated observations or a single observation should be performed, we examined whether END and transitory deterioration (TD) were associated with worse outcome, the prevalence of END and transitory deterioration (TD), and the timing of END, and whether END was detected by frequently repeated observations (Paper 2). In order to evaluate what stroke scale should be used, we compared two END definitions (Paper 3). The second main aim was to identify predictors for END (Paper 4).
We conducted a prospective observational study, where 368 patients with acute ischemic stroke were included (Protocol: Paper 1). They were observed with frequently repeated observations according to an abbreviated version of Scandinavian stroke scale (SSS), daily SSS- and NIHSS observations for the first 72 hours, and modified Rankin scale (mRS) until 90 days. Further, baseline data, risk factors and medical treatment were recorded, as well as the results from all examinations from the protocol, which was an extension of clinical routine and included non-contrast CTand/or MRI, ultrasonography or CT angiography, blood samples, electrocardiogram, heart rhythm monitoring, saturation, and blood pressure.
According to our primary definition, 13.9% had END, and 28.3% had a transitory deterioration (Paper 2). Both of these were associated with poor outcome in terms of death/dependency and length of stay in hospital. We found that our frequent observations identified all the patients with END at 72 hours. In patients with END we also found that neurological deterioration occurred throughout the first 72 hours after admission, although most of the loss of neurologica I function occurred within 24 hours. Due to the timing of END, as well as poor outcome even in patients with transitory deterioration, it is reasonable to perform frequently repeated observations, in order to detect deterioration in time for the clinician to act.
We also compared the ability of two END definitions to detect clinically relevant deterioration (Paper 3). The European Progressing Stroke Study definition (ENDEPSS), based on the Scandinavian Stroke Scale, detected END in 13.6% of patients, whereas 7.9% of patients had a ~2 point worsening in NIHSS (ENDNiHSS), i.e. 71% more with ENDEPSS (pac0.001). A poor outcome in terms of death or dependency was seen in 94 % of patients with ENDEPSS and 79% of those with ENDNIHSS, Case fatality was 24% for ENDEPSS and 21 % for ENDNIHSS In adjusted analysis, the odds ratio estimates for poor outcome was higher for ENDEPSS than ENDNIHSS, although this was not evaluated statistically. The ROC statistics indicated a closer relation with outcome (death or dependency) in the model with ENDEPSs rather than ENDNIHSS, indicating that ENDEPSS performed better. It is therefore unlikely that ENDNIHSS is superior to ENDEPSS in our setting.
In order to identify ad mission predictors for END (Paper 4), 2 patients with a peripheral cause for deterioration were defined as non-END, leaving 48 END cases. Further, patients with explanatory oedema or haemorrhagic transformation were excluded from analyses of unexplained END, leaving 35 END cases forth is substudy. In multivariable analysis, increased odds for END was seen with clinical-ASPECTS mismatch, high potassium, cardioembolic stroke, nitrate treatment, and increasing glucose, while reduced odds for END was seen for ACEI/ARB-treatment. Borderline significance for increasing odds was seen with increasing ad mission systolic blood pressure. Except nitrate treatment, aII of these were also predictors for END unexplained by oedema or haemorrhagic transformation, although the trend for ACEI/ARB-treatment was no longer significant.
These predictors partly confirm previously published results and mechanisms, while the role of high potassium, angiotensin converting enzyme in hibitor/-receptor blocker treatment and cardioembolic strokes are partly conflicting with previous findings, and partly new findings, and need confirmation in further studies. Additionally, other predictors, as blood pressure or glucose changes, should be evaluated. In conclusion, this thesis demonstrates the rationale for frequently repeated observations using key Scandinavian stroke scale items, and that it is feasible to detect deterioration in time for clinical intervention to be relevant. Also, ENDEPSS is at least as able as ENDNIHSS to detect clinically relevant deterioration at 72 hours. Additionally, the set of admission predictors identified in our study may be helpful in order to identify patients at risk. However, the impact of homeostatic changes remains to be studied.||nb_NO